Well as others have said this is an interesting article, but it is far from complete. As an owner of a P5B Deluxe, Conroe 6300, and that "expensive" OCZ ram myself, I am not able to overclock anywhere NEAR what this article claims with my "retail" B2 stepping Conroe.
You should definately post full system specs including what power supply was used, what video card, all bios settings, hard drive, voltages, tweaks, exact brand/type/config of memory, etc etc. Most AT articles have a "Test Setup" page that describes this, but this article does not.
Also I don't think "SuperPI" is a good indicator of 100% system stablity. I can overclock pretty high like 400MHz FSB too, to get my 6300 upto 2.8GHz, boot into Windows XP, and SuperPI passes, but Prime95 fails.
The highest completely stable overclock I can get on my P5B Deluxe system is 325x7 = 2.275 GHz with the 6300 Conroe. That is hours and hours of Prime95 passing.
This is a good article, just needs to be completed.
Hey all, this is probably the wrong place to post this, but is the 975X chipset going to be upgraded to allow multiplier unlocking?
Most importantly though, does anyone know if the 7950 G
X2 video card will work with the P5W-DH Deluxe? I called asus and they said they dont know lol... does anyone here have HANDS on experience with a 7950 gx2 working with a P5W-DH Deluxe?
Why tehere is no info about what memory modules were used and at what parameters.
I've read here thst i t was some superxpensive OCZ mem.
If it's true what sense is buying mainstream mobo, cheaper model of CPU and Superexpensive memory?
Anandtech has done an excellent job showing what is possible with standard cooling solutions, but I'm curious as to how high these chips could really be cranked (i.e. with water cooling). Perhaps only the EE could really benefit from this?
Provided the price keeps tumbling on these little beauties and their associated parts (mobo/ram/etc.), I think I'll be putting together a new high-end desktop with an E6600 (or higher, depends on the price) and I'd like to do some water cooling for a few reasons, mainly:
1.) noise level
2.) stability/overclockability (did I just make up a word?)
3.) it's f-ing cool (no pun intended)
If I'm understanding correctly, lowering the multiplier and increasing the bus speed should produce a much faster system overall compared to just increasing the multiplier as you're also increasing your memory speed right? So a system overclocked to say 3ghz (for number's sake) using a lower multiplier and higher bus speed would be faster than a system overclocked to 3ghz using only a higher multiplier?
And on chips that unlock "up" for the multiplier, you could still increase the bus speed for even greater performance right?
Thanks for any info. Keep up the good work Anandtech!
Yeah, but going from old school thought. higher FSBs can also mean parts wear out faster (namely the motherboard), I dont know if with memory dividers etc, if this is still the case. Anyhow, this is why I would like to see a detailed overclocking article, these little 'how-to's' all over web, and forums are in-complete, and dont offer really any insight on how this all plays a factor. However, to answer your question, YES, using a lower multiplier, with a higher front side bus, should increase performance, and I can especially see how this could play a factor if you were able to run a 1:1 memory divider.
Thanks Wesley for the info on running the memory in spec. That is exactly what I was looking for.
When I am asking for benchmarks, I would love to see PI and a few others run at the same clock speed but different FSB to show what gain we get from this..for example
The 6300 at the following:
7x429=3000
6x500=3000
What type of gain to we get from jacking up the front side bus? I know it used to help a lot on my old Barton's, but what does it get us with the Core 2?
Great article and interesting discussion. One of the things I truly love it the fact that the staff at AT respond to these post and realy add a lot to the discussion.
Thanks again!
I have a few questions that were not covered
1) Can we have some benchmarks showing the benefit of running up the FSB vs standard at the same clock speed. I know on some processors in the past this gave a nice boost in performance, but how much are we talking here? Huge gains or just a few percent?
2) Can we lock the memory on these boards and crank the FSB up with cheap memory? If so, what are the benifits the same?
Thanks for the article. I too got excited at seeing the title, hoping for a return to the golden days of overclocking without multiplier locks....
DDR2-533 is the 1:1 strap and theoretically provides the highest memory bandwidth. DDR2-533 memory setting hits DDR2-800 at 400FSB. Almost all of the cheap memory based on Elpida chips can do DDR2-800 4-3-3 as you saw in the Conroe Buying Guide. At the highest multiplier of x7, then the E6300 maxes out at 2.8GHz, with plenty of headroom left.
Dropping the memory to DDR2-400 you would reach DDR2-800 at a FSB of 533, which covers all the headroom we found with the E6300. You have even more flexibility with the E6400 with the higher x8 multiplier.
The point is you should be able to reach pretty high in headroom even with cheaper memory. Of course the high-end DDR2 all reached DDR2-1067 to DDR2-1100, which gives even more flexibility for maxing headroom at the fastest memory timings possible.
As for benchmarks, we provided Super Pi times in several of the higher overclocks as a quick bench, You can easily compare these results to what you achieve. Super Pi is a free program that's easy to run. Version 1.5 is a modified version for benchmarking available at xtremesystems.org.
We are talking about some pretty huge gains here. Just look at the new speeds, and waht the equivalent cpu would cost running at those speeds. You gain the same performance of cpu's which can cost hundreds of dollars more.
I agree wiuth your second question. I would like to see how things work out using a memory divider (do Conroe motherboards even use memory dividers?).
In "Conroe Buying Guide: Feeding the Monster" Deluxe model was featured, here we have (I suppose) P5B regular. P5B non-dlx have less overclocking options (lika limited memory voltages), this have serious impact on o/c possibilities.
In new BIOS we stilla have just 2.1 DDR2 max voltage?
New records for FSB is interesting, but I think the question that many of us have is: Is there an opportunity for getting more bang for the buck?
Assuming that upwards unlocked chips are rare, the budget-constrained conroe buyer can
a.) Crank a cheaper CPU to high FSB, with expensive RAM and motherboard. Drop the multiplier if needed to take the system to its top performance, or
b.) Save money by buying value RAM and a mid-range motherboard, and spend that extra budget on a CPU with a higher max multiplier. The max FSB will not be so exciting, and the memory may have to run at a slower ratio, but the CPU core should be quite high.
Which of these two schools of thought will perform the best for the price? Of course it will vary on a case-by-case basis, but an investigation would be very relevant.
My guess is that the $300 or so required to get a high FSB mobo and fancy RAM would be better spent on a higher multiplier. Does anyone have any thoughts on this?
(Finally, I would also like to throw in my request for a review of the Gigabyte P965-DS3, as it seems to be the cheapest available Core2 motherboard with real OC potential.)
That's very good question. An answer would depend on what is your box meant for.
It appears though that most of the applications are more "memory hungry" than CPU limited. That's why FSB speed is very important to overclockers.
quote: does the non-deluxe p5b get the same oc results?
No, expect around 7x430 with a very good setup and 7x450 with an excellent setup that is stable with the E6300. The main difference between the boards is the P5B-D has eight phase power compared to three on the P5B along with the Asus heat pipe cooling technology. The P5B-D has upgraded audio, components, additional x16 PCIE slot (x2 or x4 operation), dual Gigabit LAN, and a USB WiFi option along with additional SATA ports.
How about an article for those of us who havent OC'd in a while ? Things to test, things to change in the BIOS etc? Last truely sucsessfull OC I've personally had, was a P55 233mmx intel CPU, on a ABIT board (yes, YEARS AGO).
Anyhow, whats the impact on a system with a dropped multiplier, and increased FSB ? I'm pretty sure the longevity of said system wouldnt be as long if run stock . . .
I've got a P5B non deluxe with a E6600 + 2GB of Crucial Ballistix PC6400.
As I am no expert in overclocking the only thing I tried to up was the FSB, from 266 I managed to push it to 310, before that it would crash.
Can anybody tell me settings for say going from 2.4 to 3.0 ?
My guess is :
FSB to 333 so core speed is 333*9=3ghz
Up CPU voltage to 1.4v
..and the thing I am really not sure is the memory setting, 1. whether or not to use spd and 2. what speed to use and maybe 3. what volt to use (the Crucial is guaranteed up to 2.2v but board is limited to 2.1v).
quote: As I am no expert in overclocking the only thing I tried to up was the FSB, from 266 I managed to push it to 310, before that it would crash.
Have you tried the 0309 bios yet? It improves overclocking a great deal but comes at the expense of needing "good" memory capability in order to do it. Start with SPD, raise your memory to 2.1V, use the 1:1 divider, and set your CPU vCore at 1.375V, MCH around 1.45V as a start.
I also have the P5B non-deluxe using the new 0309 BIOS.
Core 2 Duo E6600
ASUS P5B
4GB (4x1GB) Corsair PC2-6400 4-4-4-12
I wasn't looking for max overclock as this chip is so fast already so I started by turning my multiplier down to 6, FSB to 400, and memory to 1:1. Worked fine without any voltage tweaks so I started upping my multiplier. I'm now running at 8 x 400(3.2Ghz) without messing with any voltage settings. I may be able to go higher but this is more than enough to drive the single video card that this board supports and I'm quite happy.
A weird thing I noticed is that checking the ASUS site again today(Mon. 8/28) no longer has the 0309 BIOS listed. Are some poeple having issues with it or did Intel make them remove it?
I seem to remeber hearing that the Conroes would be unlocked for all grades not just the extremes but intel was going to wait to do this for a little while after launch. it seems like I heard this even before they were called Core 2 duo's. maybe Intel locked the early one so that people would have a reason to buy the Extremes, I also seem to remember the extremes would eventually get hyperthreading back, maybe as a middleground between Kentsfield and Core 2 Duo? maybe this is why nvidia is putting off releasing any intel edition 590 boards because they didn't was to spend so much money on constantly having to update BIOS and this and that while the dust settles from the launch. not to mention the reputation. If they could release the Intel 590's without any Major bugs then that would make the new "oddball" chipset maker major kudos, at least compared to the Intel and Ati chipsets.
There's a difference between being 'in the industry' and 'clobbering people over the head with my obvious superiority and subtle hints'. If people think you're coming off as arrogant, maybe you should reconsider your approach to criticism.
I know quite a few people who know to some very important and NDA-protected facts. None of them troll the net bashing people over the head with stuff like this, because it can get you into serious trouble.
Secondly, you never never never ever try and come across as if you're speaking for the company.
Ever. Never ever. Is your supposedly leet job worth you seeming like a big man on a message board?
If you're meant to be someone of importance, you're a rank amateur at keeping your mouth shut, and a liability to boot.
My posts were for Wesley and Gary only. They will know exactly what I was talking about with the subtle hints. Those of you who feel left out in the cold by my posts, I appologize. It was never meant to be a general post.
And just to clarify, I do not speak for Intel. And again, for the record, I do not work for Intel.
This is an article comment, not a message board. The posts I made were directed to the author.
"This is an article comment, not a message board. The posts I made were directed to the author."
The author's email address is posted at the top of every page of the article. If your comment was only intended for the author, email is a much better method of communication.
"This is an article comment, not a message board. The posts I made were directed to the author."
This is where you’re wrong... the first post of a topic is generally related to the news/article but others are free and will most of the time comment your post, just like in any other message board. Even you have more than once posted a comment to posts made by other folks.
hence I think the reason for the other poster getting hostile and doubting legitimacy.
either you have information and share it or zip up...
analogous to say I overclocked my cpu to 100ghz, using 2 "special" tools, but I cant tell you what they are.
I wish Anandtech were more forthcoming with what kind of overclocks you can see with different memory/dividers, and compare the performance. Not everyone has the economical predisposition to buy super expensive "enthusiast" memory; and definitely not wanting to pay twice the price for 2-3% improvement in theoretical benchmarks, like in the past. I'm sick of the memory tunnel vision on hardware review sites.
quote: I wish Anandtech were more forthcoming with what kind of overclocks you can see with different memory/dividers, and compare the performance. Not everyone has the economical predisposition to buy super expensive "enthusiast" memory; and definitely not wanting to pay twice the price for 2-3% improvement in theoretical benchmarks, like in the past. I'm sick of the memory tunnel vision on hardware review sites.
These particular results were completed at 1:1 ratios with "very expensive" OCZ memory and is one of the major reasons why the FSB clocks are so high with the E6300/6400 chips. We realize that not everyone can afford this type of memory which was one of the main reasons behind our recent memory looks in the first Conroe buyer's guide and the DDR/DDR2 memory analysis on the ASRock boards. However, this article was about what is possible with the P5B using the top components available. As for benchmarks showing the differences between overclocks with different memory and dividers, that is a very good suggestion we will take under consideration. Thanks for the comments.
Yeah, I heard some rumours about the 965 chipset not overclocking well with a lower than 1:1 ration on the ram.
I am interested in a low-cost overclock with 667mhz ddr2 ram and the E6300, to get the most bang for the buck. - Actually to use some of the money saved on the gfx card instead, due to most games beeing gfx card limited.
In other words, a review of the P5B vanilla version with a E6300 and some generic ddr2 667 mhz ram would be really interesting ;)
Thanks, it's good to know that you'll consider such an article. The buyer guide article was good for the time yes. My reasoning is that it kind of defeats the purpose to buy the cheapest Core 2 Duo when you have to buy the most expensive memory to fully utilize it. Makes sense, I hope.
I suppose I could be glib and comment that Asus and Gigabyte will soon be the same company so our original reporting is correct. But I won't do that :) Several weeks ago it was announced that the two companies would merge, but brand identities will likely remain distinct - at least for a while.
Gary and I did have a discussion about the Gigabyte board this afternoon, but there was a misunderstading between us about the unlock feature on the Gigabyte. Gary has explained this in detail in comments above, and he revised the article to reflect that the Gigabyte board also unlocks down, since not mentioning the Gigabyte was clearly an unintentional error that is now corrected. That will teach Gary to set me up :) Actually, Gary and I have been friends for years - long before I was with AT or Gary came on board with us. Gary is incredibly conscientious, and we are fortunate to have him on staff.
I just got off the phone with Asus Engineering (it's Friday morning in Taiwan) who will provide more information on the chips they claim work on the up unlock and more information on how it works. ALL of the chips we tested unlocked down. None of the chips at AT - "Cherry-Picked", retail, E6700, E6600, E6400, E6300 would unlock and clock up - as we reported CLEARLY. We will update the article with the additional info from Asus when we get it.
As for nothing new, I certainly consider the overclcoks we got pretty exciting. Gary owns the highest overclocks at 525 and 532 on several Forums since he unlocked the P5B. It is now clear the Gigabyte has the unlock down feature also, but I know for a fact the record overclocks didn't happen on the Gigabyte.
oh, not exactly a "merger"
more like a partnership
they will be producing a new set of boards together ala asrock type boards?
any news if this partnership will be producing budget boards or enthusiast type boards, the press release doesnt mention that.
has a name been decided yet?
giga-sus ? lol
as-byte? double lol
I suspect this will be more a asrock type venture, where cheap mass produced boards can be made without affecting the parent brand name if questions of quality arise
the benefits though as with asrock, have been the company being able to be more adventurous on designs and configurations?
When I first saw the title of this article I was extremely excited.I was thinking to myself: "Asus adds unlocking? Awesome! I'll finally be able unlock those multipliers upwards on some cheap Conroes!" I have a feeling a lot of other people had the same initial enthusiasm when they saw the title.
Of course then I got to the point where it's mentioned that none of the Conroes unlocked upwards. Given that the technology mentioned in this article has existed for some time, I'm not sure why the article was even put on Anandtech.
Exactly. The whole article has been singing in ASUS tune when in fact they've just copied someone else's feature! Shame shame.
It's such an old feature to exploit EIST for multiplier adjustment that Gigabyte didn't even talked about it any more. Perhaps Wesley is excited ASUS has finally implemented that feature? :) ASUS fan-boy.
quote: Exactly. The whole article has been singing in ASUS tune when in fact they've just copied someone else's feature! Shame shame.
This is my fault. I had a miscommunication with Wes about the unlock features on the Gigabyte boards. Asus had figured out a way to go up and down which (we are trying to find the proper lot number to test) Gigabyte and others had not at this time. In my conversations with Wes I relayed the wrong statement about Gigabyte's capability in this area or my words were confusing about the EIST exploit. Asus will also implement this capability on the 975X chipset in the near future. If the tables were turned on the manufacturer then Wes would be a Gigabyte Fan-Boy I guess......
Find a bigger sample -- determine which ones can be "unlocked" upwards, which ones can't. Compare with Retail B2. Those are the ones most people are using and are available off the shelves. Otherwise, the article is completely meaningless as it is right now. Other than to make Wesley appear like an ASUS fan-boy, of course! ;)
Anandtech articles are so significant to a lot of people and companies -- you simply can't afford to make such elementary mistakes! If some motherboard manufacturers tell you they have unlocked one of Intel's most highly kept secrets, you'll have to start questioning from every angle -- not embrace it like every other fan-boys.
Like I've said, there're only 2 tools that are capable of 'unlocking' (it's actually reversing) the fused settings. 3 if you consider micro surgery on the chip itself. Never had the need to use that big expensive machine though. If the chip is properly configured (i.e. all production parts), you can forget reversing it. I can't go into more details about the 2 tools, but I can assure you that ASUS or any other mobo manufacturers for that matter do not have access to them.
quote: Anandtech articles are so significant to a lot of people and companies -- you simply can't afford to make such elementary mistakes! If some motherboard manufacturers tell you they have unlocked one of Intel's most highly kept secrets, you'll have to start questioning from every angle -- not embrace it like every other fan-boys.
We agree. :) We do have screenshots from Asus along with other engineering information showing chips upclocking. We are discussing the upclock statements but for the mean time the article has clarified the down clock information. That error was my fault as my communication to Wes was flawed when discussing the down clock with and without EIST enabled.
As for the B1 / B2 steppings, B1 was the planned retail release and several shipments did make it into distribution. Our B2 E6300/E6400 overclock better than their B1 counterparts. As for the B1 E6600~X6800, our internal units vary up to 200MHZ in overclock ability on the X6800 as an example. Intel sent out several thousand B1 review samples and that is what we have to use unless individuals on the staff purchase new retail chips which can become quite expensive. ;-) I have been through all of the overclocking results with our press samples/retail chips. Unless you had access to hand binned chips, then it truly is the luck of the draw about the chip capability you will receive.
Zach, for someone who's not supposed to say much, you talk a lot... and not by email, which I find surprising. It's your call, but it is very unusual, and motivated by something I think.
They never said they unlocked up. And GIVE THE B2 vs. B1 SHIT A REST!
Did they say their B1 unlocked up? NO! Did they say their B2 unlocked up? NO! So what exactly does B2 give you over B1, a display in CPU-Z that saying B2 and probably a processor that you didn't have to pay for, that's what. ASUS says some chips do unlock up; I certainly don't expect AnandTech to run around purchasing chips to try and find out which ones will work.
So Gigabyte and ASUS unlock down, other boards use EIST to automatically drop to 6X (and only 6X as far as I'm aware) for power saving. BIOS updates can fix that, but a lot of companies won't bother. Anyway, at least we now know that you work for Intel, what with your L33T Inside Information.
Thanks, old Intel fart who probably works in Malaysia for Intel at a chip fab. As I said, never say never. If there's one thing I can say for certain, it's that the Taiwanese, Chinese, etc. have some very talented engineers. I don't doubt that Asus has some non-ES, non-X6800 chips that unlock up. Are they simply an analomy? Possibly - probably even - but you "2 sooper secrit toolz" that allow reversal of the locks simply mean that there is a way to pull this off. Whether it's practical or not, I can't say, and you're not telling.
Anyway, they've updated the conclusion to now mention Gigabyte, so I hope that keeps you happy. I for one must also be an Asus fanboy because I find their Intel boards to be among the best around - far better than the BadAxe, that's for sure. Without knowing anything about a motherboard, I'd go with chance and take an Asus over any other manufacturer. There have been a few lemons, but everyone has those - even Intel (and AMD, Gigabyte, etc.) By and large, I trust Asus more than any other motherboard company.
And hey, if they want to market their boards better than Gigabyte and specifically ask Anandtech to do a special report on the feature, more power to 'em! AMD could learn a thing or two about marketing from Asus. Heh. I still prefer the P5W DH, but would really like to see SLI support with a high-end chipset. (Yeah, I'm an nVidia fanboy as well - not my fault that ATI's drivers are still flaky in a lot of situations and have a cruddy UI.) Unfortunately, I don't think 590 SLI is going to be the answer... maybe "nforce6" or whatever will offer high overclocks and SLI support, but that's still a while off it seems.
And no, you're way off the point -- this is never about Gigabyte vs ASUS.
And yes, you're entitled to have your own opinion. Perhaps one day we'll even have a mobo that allows you to have both Intel and AMD chips on one board doing dual processor. Great, now off you go.
quote: There's no way you can unlock it upwards -- unless you have the 2 tools I had access to. And no, you do not have access to those tools.
quote: Like I've said, there're only 2 tools that are capable of 'unlocking' (it's actually reversing) the fused settings. 3 if you consider micro surgery on the chip itself. Never had the need to use that big expensive machine though. If the chip is properly configured (i.e. all production parts), you can forget reversing it. I can't go into more details about the 2 tools, but I can assure you that ASUS or any other mobo manufacturers for that matter do not have access to them.
Okay, so you don't work for Intel but you apparently have access to a microsurgery chip corrector machine, and you have access to tools that "ASUS or any other mobo manufacturers" lack. Wow. If you don't work for Intel, then I would have to conclude you either work for AMD, IBM, or some other major chip fabrication company (TSMC maybe). Then we of course have to wonder why this upsets you so... but if you work for AMD that would be a given, since good news for Intel is bad news for AMD generally speaking. I have difficulty imagining something that you might have access to that Asus would not.
I suppose you need to protect your identity, regardless, but it's all rather silly to bash a short article that basically says "Asus claims upward unlocks are sometimes possible" when you really don't have proof, except from your vast knowledge of what is and isn't possible. The world is full of things that were once deemed "impossible", and if Anandtech says ASUS has sent them screencaptures, I have no reason to doubt that. Were they with ES, B1, B2, or something else? Nobody has yet given a concrete answer.
Do I trust you, or Anandtech? I forgot, AT is full of Intel, AMD, Asus, Gigabyte, ASRock, etc. fanboys, so I guess we should all trust Zach instead. Great, now off you go. Hope they pay you well. Goodnight.
There's no way you can unlock it upwards -- unless you have the 2 tools I had access to. And no, you do not have access to those tools.
The only way ASUS managed to 'unlock' is simply because Intel did not fuse out the multiplier in certain engineering samples. There're different classes of samples. I am not allowed to go into too much details, but the article is simply wrong.
Anyway, if he sang so much praises for Gigabyte in one article, yeah, I'll call him a Gigabyte fan-boy. :) Since he can't proved that ASUS unlocked it upwards on Retail processors, he ignorantly praised ASUS nevertheless. You'd have to agree that it warrants the label of fan-boy! ;)
Only thing I can say for certain is that you're coming off as an arrogant prick, Zach. I sure hope you work for a motherboard/RAM/CPU manufacturer. As for what is and isn't possible with BIOS tweaks, never say never. Maybe Intel just had a batch that didn't get locked properly, maybe not.
But a FSB overclock to 532 MHz is definitely nothing to scoff at, and whether Gigabyte was first with a buggy BIOS or not doesn't matter much to me. After all, the ftp://dlsvr01.asus.com/pub/ASUS/mb/socket775/P5B%2...">0507 Asus BIOS is from 2006/8/11 which makes it two weeks old. That's about the time Core 2 really became available anyway. Gigabyte's F4 came out around the same time, and that's the first truly usable BIOS from them as far as I can tell.
quote: Only thing I can say for certain is that you're coming off as an arrogant prick, Zach.
If you're in my position, knowing so much without the liberty of sharing the knowledge, you'll know how I feel. So it's not arrogance. I simply am not allowed to disclose too much. And do feel free to continue speculating. I've allowed you guys to go on for years without intervention, but this simply crosses the line. Anyhow, you probably won't be hearing much from me any more. If I've appeared to you as an arrogant prick trying to guide the authors to the correct path, then so be it.
I agree with you. I'm sure there is absolutely no way to change the multiplier UPWARDS on a stock E6300-E6700. Given the overclocking potential of any of these CPUs, everyone could simply make his cheaper CPU a X6800 (or X6800 with 2MB cache :) And this really is the last thing Intel wants.
The article wasn't really so bad as the criticisms seem to tell, imo. Abusing IST has been around a while I admit but the basis of a stronger memory controller bios still stands and this board is significantly stronger than Gigabyte's in that regard alone.
There were B1 steppings that made it to retail, at least according to a few of the recording threads I've seen.
I think the only thing truly sad is that the bios's have been abhorrently buggy on the 975's the 965's and the 590's. People have started to measure which is the "least buggy" which is kind of crazy at the prices these boards are commanding. People need to let these things cool a bit, and mature a LOT, before they're going to be worth what's being asked for them. Only after things have matured and are tested are we really going to see which of all is worthy of the money. And sadly, from what I'm hearing, the 590 is definitely NOT going to be it, so better keep an eye on the other two.
quote: There were B1 steppings that made it to retail, at least according to a few of the recording threads I've seen.
Nope.
quote: he basis of a stronger memory controller bios still stands and this board is significantly stronger than Gigabyte's in that regard alone.
It's true that Gigabyte sucks when it comes to memory compatibility. But this article is not about that, is it?
If Gigabyte fixes the issues on memory compatibility, and the author writes about it saying that it's the first manufacturer to come out with a bios that is highly compatible with a lot of memory modules, I'll still be ranting about it.
Well, perhaps the author shouldn't jump at the very first chance to suck the toe of ASUS' BIOS programmer... check around forums to find out if other manufacturers have this feature as well. A little research is always required before you start writing an article that will be read by thousands of people. It's the author's obligation.
If ASUS has added this feature for the P5B does that mean that the other boards for Core 2 (most importantly for me the nForce 500's) will have this feature?
Why's everyone so worried about other boards not having it?
It's a very simple addition. Gimme the BIOS source, I'll add it in for u. It's a default 'feature' of the CPU, so whether or not you get it in your board will depend on how diligent your mobo's BIOS programmer is. Shouldn't take longer than 30 minutes to implement and test (internally of course). It's the process of complete validation of a new release that most BIOS programmers fear.
"the ability to unlock the previously locked Conroe chips is so significant that every other manufacturer will have to follow the ASUS lead or lose market share to the motherboard giant."
LOL. Perhaps ASUS is the one who is the follower. All of Gigabyte's 965 boards have had this ability since the very beginning. ASUS had to release an update to their BIOS to support this, after seeing Gigabyte's feature.
In any event, this is a VERY SIMPLE function to implement -- what innovation? It simply uses the EIST feature in your CPU to implement multiplier "unlock". In fact, it's technically wrong to say "unlock", since we don't lock the multiplier you managed to select in the first place! That's the whole point of EIST. Unless you can select upwards, then you can claim to have unlocked the multiplier. Otherwise, it's just legally and technically a wrong term to use. You can claim to support Multiplier Adjustment, but do not EVER say it's unlocked. Be warned.
Please clean up your article. Too many mistakes.
Also, it'll be good to put a foreword in your article to mention that your oc is based on Engineering Samples (B1 step), which are different from retail (B2). Retail ones have some differences which make it a little faster clock to clock, but not as overclockable. You do not want people to rush out to buy those processors, based on your table of overclocked frequencies.
Yeah, "Golden Cherry Picked units" like the freaking RETAIL E6400 and E6300 stepping B2 that overclocked higher than the "cherry" ES B1 E6600/E6700? OMFG the world is coming to an end!!!
The only thing potentially misleading is that the B2 E66/6700 might do better than shown here. The bottom line is that the B2 is a later, IMPROVED stepping that should actually at least match these results.
The Gigabyte information is more of a problem, but given my experience with a DS3 I can't say I'm too happy with GB right now. F4 seems to have cleared up most of the problems. F2 was an absolute joke, however.
quote: The bottom line is that the B2 is a later, IMPROVED stepping that should actually at least match these results.
You simply do not know enough. B2 is a later step, no doubt. But, who says later steps increase oc-potential??? It's pretty weird how the community came to this "conclusion". It might be true for later steps (C and above), but usually not the earlier ones such as B1 and B2. So don't jump to conclusions when you have never tested out A0 parts before.
All I'm asking them to do is, to use retail parts. That's all.
Maybe you can pull some strings with Intel and get them to send new B2 or later retail samples to Anandtech? I'm sure the AT guys don't want to go out and buy a bunch of new CPUs when the current ones work fine. We all recognize that there is an element of marketing to hardware enthusiast sites - that's why there are a shitload of ads at all the sites, right? (Thank you, AdBlock!) AMD and Intel (and Asus, Gigabyte, Asrock, etc.) send parts for free to AT and they review them. "No such thing as bad publicity!"
You keep ignoring the fact that the E63/6400 were NOT B1/ES steppings, and yet they overclocked better. That's where I drew my conclusions. Sure, less cache has an impact, but at the very least it's safe to say B1 isn't amazingly better than B2. Anyway, all overclocks are largely CPU dependent. Everyone in the community knows that. YMMV: Your Mileage May Vary. That the P5B hits 500+ MHz FSB is more than enough for me. I don't particularly like to run at "mostly stable" settings, so if I can POST and run most benches at a 68% OC, I'll drop to 50% for 100% stability and peace of mind. $185-$200 for a chip that can match stock X6800 performance is pretty tasty.
1) How is it possible that they can do this? I thought the multiplier lock was something phyiscally hard-wired in the manufacture of the processor - akin to those graphics cards that absolutely cannot unlock extra pipes.
2) Any chance that we'll see it appearing on the other ASUS conroe boards? Or is the answer to (1) intricately linked to the north bridge?
3) How about other manufacturers? Now that ASUS has set the stage, are we going to see the Gigabyte (or what's left of 'em), MSI, DFI, etc... boards following suit? Or is this patentable technology?
4) Anyone else had it with these <explicative> snakes on this <explicative> plane? I wish someone would do something about them, already....
quote: 1) How is it possible that they can do this? I thought the multiplier lock was something phyiscally hard-wired in the manufacture of the processor - akin to those graphics cards that absolutely cannot unlock extra pipes.
Write a few bytes into the MSR. Read the PRM of conroe. You'll find out how easy it actually is. Under Windows, you'll need a driver's help to access MSRs, but the BIOS has complete access to MSRs.
quote: 2) Any chance that we'll see it appearing on the other ASUS conroe boards? Or is the answer to (1) intricately linked to the north bridge?
ASUS followed Gigabyte -- AFAIK, Gigabyte already has this feature since they introduced 965P-series boards. They did it so early that they didn't need any updates.
quote: ASUS followed Gigabyte -- AFAIK, Gigabyte already has this feature since they introduced 965P-series boards. They did it so early that they didn't need any updates.
I can confirm this. I just tried it on my DS3 with and E6400. I was able to set the multi to 6, 7, and 8.
We agree with you. We stated what we found - all Conroes we tested unlocked down. We also told you what Asus told us, but we were clear it could not be confirmed in our own testing. We did see screen shots of a locked Conroe at higher multipliers, but we did not find one in our testing. Our coverage was fair IMHO.
I would say the coverage was helpful, but not fair. Or if it was fair, it wasn't what I except from AT. I think normally you guys are pretty tough on campanies that don't deliver on promises, which is great because:
-You're an advocate for the consumer
-You serve as a counterweight to huge manufacturer-funded marketing efforts.
I think your position on promised video-card launch dates that came and went was more what I like - you reported the manufacturer's stated date, but also blasted the GPU maker for letting the dates slip in the past and suggested it was likely to happen again.
It's fine to publish ASUS's claims, after all you want them to send you products to review, but why not editorialize a little more. Not have one chip that did what they claimed was sure plenty of justification.
So we should blame ASUS for the lack of Conroe chips at AnandTech that can be unlocked upwards? AT stated ASUS' claims and warned that they weren't able to verify all of them. It sounds like you just want the authors of the article to bash someone.
I want the authors to keep up working on behalf of the consumer. When a company touts a feature that isn't present in many if not most cases, the company shouldn't get much credit for the "feature". Repeating the party line offered by ASUS gives them more credibility than they need to be given here. You and I can all go to ASUS's site and read up on their marketing department's list of reasons to buy their board, I expect AT to revivew the board with an eye to using their expertiese to inform me about it's strengths and weaknesses. An advertised feature that is present in ZERO percent of tested boards is more of a minus than a plus.
We’ve updated our terms. By continuing to use the site and/or by logging into your account, you agree to the Site’s updated Terms of Use and Privacy Policy.
84 Comments
Back to Article
HammerD - Wednesday, September 6, 2006 - link
Well as others have said this is an interesting article, but it is far from complete. As an owner of a P5B Deluxe, Conroe 6300, and that "expensive" OCZ ram myself, I am not able to overclock anywhere NEAR what this article claims with my "retail" B2 stepping Conroe.You should definately post full system specs including what power supply was used, what video card, all bios settings, hard drive, voltages, tweaks, exact brand/type/config of memory, etc etc. Most AT articles have a "Test Setup" page that describes this, but this article does not.
Also I don't think "SuperPI" is a good indicator of 100% system stablity. I can overclock pretty high like 400MHz FSB too, to get my 6300 upto 2.8GHz, boot into Windows XP, and SuperPI passes, but Prime95 fails.
The highest completely stable overclock I can get on my P5B Deluxe system is 325x7 = 2.275 GHz with the 6300 Conroe. That is hours and hours of Prime95 passing.
This is a good article, just needs to be completed.
isvaljek - Sunday, September 3, 2006 - link
Does this mean E6300 can run at 3.5GHz with FSB:RAM 1:1 with plain 533MHz PC4200 RAM?LordX2 - Wednesday, August 30, 2006 - link
Hey all, this is probably the wrong place to post this, but is the 975X chipset going to be upgraded to allow multiplier unlocking?Most importantly though, does anyone know if the 7950 G
X2 video card will work with the P5W-DH Deluxe? I called asus and they said they dont know lol... does anyone here have HANDS on experience with a 7950 gx2 working with a P5W-DH Deluxe?
Any info or links would be greatly appreciated!
JyriT - Wednesday, August 30, 2006 - link
I've got the P5B, E6400 and DDR2 800. Does it sound like a good idea to set the memory/FSB for 400 MHz and processor for 6x400 MHz?KHysiek - Sunday, August 27, 2006 - link
Why tehere is no info about what memory modules were used and at what parameters.I've read here thst i t was some superxpensive OCZ mem.
If it's true what sense is buying mainstream mobo, cheaper model of CPU and Superexpensive memory?
gilahacker - Saturday, August 26, 2006 - link
Anandtech has done an excellent job showing what is possible with standard cooling solutions, but I'm curious as to how high these chips could really be cranked (i.e. with water cooling). Perhaps only the EE could really benefit from this?Provided the price keeps tumbling on these little beauties and their associated parts (mobo/ram/etc.), I think I'll be putting together a new high-end desktop with an E6600 (or higher, depends on the price) and I'd like to do some water cooling for a few reasons, mainly:
1.) noise level
2.) stability/overclockability (did I just make up a word?)
3.) it's f-ing cool (no pun intended)
If I'm understanding correctly, lowering the multiplier and increasing the bus speed should produce a much faster system overall compared to just increasing the multiplier as you're also increasing your memory speed right? So a system overclocked to say 3ghz (for number's sake) using a lower multiplier and higher bus speed would be faster than a system overclocked to 3ghz using only a higher multiplier?
And on chips that unlock "up" for the multiplier, you could still increase the bus speed for even greater performance right?
Thanks for any info. Keep up the good work Anandtech!
yyrkoon - Saturday, August 26, 2006 - link
Yeah, but going from old school thought. higher FSBs can also mean parts wear out faster (namely the motherboard), I dont know if with memory dividers etc, if this is still the case. Anyhow, this is why I would like to see a detailed overclocking article, these little 'how-to's' all over web, and forums are in-complete, and dont offer really any insight on how this all plays a factor. However, to answer your question, YES, using a lower multiplier, with a higher front side bus, should increase performance, and I can especially see how this could play a factor if you were able to run a 1:1 memory divider.yyrkoon - Saturday, August 26, 2006 - link
err sorry, meant to say lower multiplier with a higher FSB should improve performance, as long as the CPU speed remains the same.zemane - Saturday, August 26, 2006 - link
Next year's BIOS (seen on page 1)Gary Key - Saturday, August 26, 2006 - link
That is why the bios is so special, it is a year ahead of its time. LOL... I will get that changed and actually the 0605 bios is out now.BuddyHolly - Friday, August 25, 2006 - link
Thanks Wesley for the info on running the memory in spec. That is exactly what I was looking for.When I am asking for benchmarks, I would love to see PI and a few others run at the same clock speed but different FSB to show what gain we get from this..for example
The 6300 at the following:
7x429=3000
6x500=3000
What type of gain to we get from jacking up the front side bus? I know it used to help a lot on my old Barton's, but what does it get us with the Core 2?
Great article and interesting discussion. One of the things I truly love it the fact that the staff at AT respond to these post and realy add a lot to the discussion.
Thanks again!
BuddyHolly - Friday, August 25, 2006 - link
I have a few questions that were not covered1) Can we have some benchmarks showing the benefit of running up the FSB vs standard at the same clock speed. I know on some processors in the past this gave a nice boost in performance, but how much are we talking here? Huge gains or just a few percent?
2) Can we lock the memory on these boards and crank the FSB up with cheap memory? If so, what are the benifits the same?
Thanks for the article. I too got excited at seeing the title, hoping for a return to the golden days of overclocking without multiplier locks....
Wesley Fink - Friday, August 25, 2006 - link
DDR2-533 is the 1:1 strap and theoretically provides the highest memory bandwidth. DDR2-533 memory setting hits DDR2-800 at 400FSB. Almost all of the cheap memory based on Elpida chips can do DDR2-800 4-3-3 as you saw in the Conroe Buying Guide. At the highest multiplier of x7, then the E6300 maxes out at 2.8GHz, with plenty of headroom left.Dropping the memory to DDR2-400 you would reach DDR2-800 at a FSB of 533, which covers all the headroom we found with the E6300. You have even more flexibility with the E6400 with the higher x8 multiplier.
The point is you should be able to reach pretty high in headroom even with cheaper memory. Of course the high-end DDR2 all reached DDR2-1067 to DDR2-1100, which gives even more flexibility for maxing headroom at the fastest memory timings possible.
As for benchmarks, we provided Super Pi times in several of the higher overclocks as a quick bench, You can easily compare these results to what you achieve. Super Pi is a free program that's easy to run. Version 1.5 is a modified version for benchmarking available at xtremesystems.org.
shabby - Saturday, August 26, 2006 - link
So the p5b has a ddr2-400 setting? I thought the lowest was a 1:1 ratio(533 setting).Gary Key - Saturday, August 26, 2006 - link
If you are using a P4 then the DDR2-400 setting is the base setting on this board.
shabby - Sunday, August 27, 2006 - link
And if its a c2d then your outta luck?kmmatney - Friday, August 25, 2006 - link
We are talking about some pretty huge gains here. Just look at the new speeds, and waht the equivalent cpu would cost running at those speeds. You gain the same performance of cpu's which can cost hundreds of dollars more.I agree wiuth your second question. I would like to see how things work out using a memory divider (do Conroe motherboards even use memory dividers?).
KHysiek - Friday, August 25, 2006 - link
In "Conroe Buying Guide: Feeding the Monster" Deluxe model was featured, here we have (I suppose) P5B regular. P5B non-dlx have less overclocking options (lika limited memory voltages), this have serious impact on o/c possibilities.In new BIOS we stilla have just 2.1 DDR2 max voltage?
junior1 - Friday, August 25, 2006 - link
New records for FSB is interesting, but I think the question that many of us have is: Is there an opportunity for getting more bang for the buck?Assuming that upwards unlocked chips are rare, the budget-constrained conroe buyer can
a.) Crank a cheaper CPU to high FSB, with expensive RAM and motherboard. Drop the multiplier if needed to take the system to its top performance, or
b.) Save money by buying value RAM and a mid-range motherboard, and spend that extra budget on a CPU with a higher max multiplier. The max FSB will not be so exciting, and the memory may have to run at a slower ratio, but the CPU core should be quite high.
Which of these two schools of thought will perform the best for the price? Of course it will vary on a case-by-case basis, but an investigation would be very relevant.
My guess is that the $300 or so required to get a high FSB mobo and fancy RAM would be better spent on a higher multiplier. Does anyone have any thoughts on this?
(Finally, I would also like to throw in my request for a review of the Gigabyte P965-DS3, as it seems to be the cheapest available Core2 motherboard with real OC potential.)
Janooo - Friday, August 25, 2006 - link
That's very good question. An answer would depend on what is your box meant for.It appears though that most of the applications are more "memory hungry" than CPU limited. That's why FSB speed is very important to overclockers.
Visual - Friday, August 25, 2006 - link
does the non-deluxe p5b get the same oc results?what exactly are the differences between deluxe and normal?
Gary Key - Friday, August 25, 2006 - link
No, expect around 7x430 with a very good setup and 7x450 with an excellent setup that is stable with the E6300. The main difference between the boards is the P5B-D has eight phase power compared to three on the P5B along with the Asus heat pipe cooling technology. The P5B-D has upgraded audio, components, additional x16 PCIE slot (x2 or x4 operation), dual Gigabit LAN, and a USB WiFi option along with additional SATA ports.
yyrkoon - Friday, August 25, 2006 - link
How about an article for those of us who havent OC'd in a while ? Things to test, things to change in the BIOS etc? Last truely sucsessfull OC I've personally had, was a P55 233mmx intel CPU, on a ABIT board (yes, YEARS AGO).Anyhow, whats the impact on a system with a dropped multiplier, and increased FSB ? I'm pretty sure the longevity of said system wouldnt be as long if run stock . . .
daneel3001 - Friday, August 25, 2006 - link
I've got a P5B non deluxe with a E6600 + 2GB of Crucial Ballistix PC6400.As I am no expert in overclocking the only thing I tried to up was the FSB, from 266 I managed to push it to 310, before that it would crash.
Can anybody tell me settings for say going from 2.4 to 3.0 ?
My guess is :
FSB to 333 so core speed is 333*9=3ghz
Up CPU voltage to 1.4v
..and the thing I am really not sure is the memory setting, 1. whether or not to use spd and 2. what speed to use and maybe 3. what volt to use (the Crucial is guaranteed up to 2.2v but board is limited to 2.1v).
Cheers
Dan
Gary Key - Friday, August 25, 2006 - link
Have you tried the 0309 bios yet? It improves overclocking a great deal but comes at the expense of needing "good" memory capability in order to do it. Start with SPD, raise your memory to 2.1V, use the 1:1 divider, and set your CPU vCore at 1.375V, MCH around 1.45V as a start.
daneel3001 - Sunday, August 27, 2006 - link
Yep tried the 0309.I am currently set at :
CPU vcore @ 1.4v but showing 1.368v on CPU-Z
Mem voltage @ 2.1v
FSB @ 340
Mem @ 428 (860 DDR2).
Not sure about MCH you mentioned.
For some reason I can't get my machine stable past 3ghz, SuperPi will be ok but Sandra burnin will crash. I'm going to try with current settings..
Dan
DudemanX - Monday, August 28, 2006 - link
I also have the P5B non-deluxe using the new 0309 BIOS.Core 2 Duo E6600
ASUS P5B
4GB (4x1GB) Corsair PC2-6400 4-4-4-12
I wasn't looking for max overclock as this chip is so fast already so I started by turning my multiplier down to 6, FSB to 400, and memory to 1:1. Worked fine without any voltage tweaks so I started upping my multiplier. I'm now running at 8 x 400(3.2Ghz) without messing with any voltage settings. I may be able to go higher but this is more than enough to drive the single video card that this board supports and I'm quite happy.
A weird thing I noticed is that checking the ASUS site again today(Mon. 8/28) no longer has the 0309 BIOS listed. Are some poeple having issues with it or did Intel make them remove it?
lopri - Friday, August 25, 2006 - link
I can't help but to think this article is some kind of a stunt, which is in line with AT/ASUS alliance. It'd make a good Inquirer article, though.Black69ta - Wednesday, August 30, 2006 - link
I seem to remeber hearing that the Conroes would be unlocked for all grades not just the extremes but intel was going to wait to do this for a little while after launch. it seems like I heard this even before they were called Core 2 duo's. maybe Intel locked the early one so that people would have a reason to buy the Extremes, I also seem to remember the extremes would eventually get hyperthreading back, maybe as a middleground between Kentsfield and Core 2 Duo? maybe this is why nvidia is putting off releasing any intel edition 590 boards because they didn't was to spend so much money on constantly having to update BIOS and this and that while the dust settles from the launch. not to mention the reputation. If they could release the Intel 590's without any Major bugs then that would make the new "oddball" chipset maker major kudos, at least compared to the Intel and Ati chipsets.splines - Thursday, August 24, 2006 - link
There's a difference between being 'in the industry' and 'clobbering people over the head with my obvious superiority and subtle hints'. If people think you're coming off as arrogant, maybe you should reconsider your approach to criticism.I know quite a few people who know to some very important and NDA-protected facts. None of them troll the net bashing people over the head with stuff like this, because it can get you into serious trouble.
Secondly, you never never never ever try and come across as if you're speaking for the company.
Ever. Never ever. Is your supposedly leet job worth you seeming like a big man on a message board?
If you're meant to be someone of importance, you're a rank amateur at keeping your mouth shut, and a liability to boot.
ZachSaw - Friday, August 25, 2006 - link
My posts were for Wesley and Gary only. They will know exactly what I was talking about with the subtle hints. Those of you who feel left out in the cold by my posts, I appologize. It was never meant to be a general post.And just to clarify, I do not speak for Intel. And again, for the record, I do not work for Intel.
This is an article comment, not a message board. The posts I made were directed to the author.
Questar - Friday, August 25, 2006 - link
I find that people who say in public how superior they are to others generally aren't.Araemo - Friday, August 25, 2006 - link
"This is an article comment, not a message board. The posts I made were directed to the author."The author's email address is posted at the top of every page of the article. If your comment was only intended for the author, email is a much better method of communication.
It might also get you a more direct response. ;)
Strunf - Friday, August 25, 2006 - link
"This is an article comment, not a message board. The posts I made were directed to the author."This is where you’re wrong... the first post of a topic is generally related to the news/article but others are free and will most of the time comment your post, just like in any other message board. Even you have more than once posted a comment to posts made by other folks.
splines - Thursday, August 24, 2006 - link
That was for Zachsaw, by the way :)xsilver - Thursday, August 24, 2006 - link
hence I think the reason for the other poster getting hostile and doubting legitimacy.either you have information and share it or zip up...
analogous to say I overclocked my cpu to 100ghz, using 2 "special" tools, but I cant tell you what they are.
mongo lloyd - Thursday, August 24, 2006 - link
I wish Anandtech were more forthcoming with what kind of overclocks you can see with different memory/dividers, and compare the performance. Not everyone has the economical predisposition to buy super expensive "enthusiast" memory; and definitely not wanting to pay twice the price for 2-3% improvement in theoretical benchmarks, like in the past. I'm sick of the memory tunnel vision on hardware review sites.Gary Key - Friday, August 25, 2006 - link
These particular results were completed at 1:1 ratios with "very expensive" OCZ memory and is one of the major reasons why the FSB clocks are so high with the E6300/6400 chips. We realize that not everyone can afford this type of memory which was one of the main reasons behind our recent memory looks in the first Conroe buyer's guide and the DDR/DDR2 memory analysis on the ASRock boards. However, this article was about what is possible with the P5B using the top components available. As for benchmarks showing the differences between overclocks with different memory and dividers, that is a very good suggestion we will take under consideration. Thanks for the comments.
Questar - Friday, August 25, 2006 - link
Thanks Gary, this is exactly what we need. The question is "do memory dividers make a significant difference for Conroe".blackbrrd - Sunday, August 27, 2006 - link
Yeah, I heard some rumours about the 965 chipset not overclocking well with a lower than 1:1 ration on the ram.I am interested in a low-cost overclock with 667mhz ddr2 ram and the E6300, to get the most bang for the buck. - Actually to use some of the money saved on the gfx card instead, due to most games beeing gfx card limited.
In other words, a review of the P5B vanilla version with a E6300 and some generic ddr2 667 mhz ram would be really interesting ;)
mongo lloyd - Friday, August 25, 2006 - link
Thanks, it's good to know that you'll consider such an article. The buyer guide article was good for the time yes. My reasoning is that it kind of defeats the purpose to buy the cheapest Core 2 Duo when you have to buy the most expensive memory to fully utilize it. Makes sense, I hope.Wesley Fink - Thursday, August 24, 2006 - link
I suppose I could be glib and comment that Asus and Gigabyte will soon be the same company so our original reporting is correct. But I won't do that :) Several weeks ago it was announced that the two companies would merge, but brand identities will likely remain distinct - at least for a while.Gary and I did have a discussion about the Gigabyte board this afternoon, but there was a misunderstading between us about the unlock feature on the Gigabyte. Gary has explained this in detail in comments above, and he revised the article to reflect that the Gigabyte board also unlocks down, since not mentioning the Gigabyte was clearly an unintentional error that is now corrected. That will teach Gary to set me up :) Actually, Gary and I have been friends for years - long before I was with AT or Gary came on board with us. Gary is incredibly conscientious, and we are fortunate to have him on staff.
I just got off the phone with Asus Engineering (it's Friday morning in Taiwan) who will provide more information on the chips they claim work on the up unlock and more information on how it works. ALL of the chips we tested unlocked down. None of the chips at AT - "Cherry-Picked", retail, E6700, E6600, E6400, E6300 would unlock and clock up - as we reported CLEARLY. We will update the article with the additional info from Asus when we get it.
As for nothing new, I certainly consider the overclcoks we got pretty exciting. Gary owns the highest overclocks at 525 and 532 on several Forums since he unlocked the P5B. It is now clear the Gigabyte has the unlock down feature also, but I know for a fact the record overclocks didn't happen on the Gigabyte.
xsilver - Thursday, August 24, 2006 - link
asus and gigabyte merge?this is news to me.. link?
this would cause some serious monopolization of the market, which cant be a good thing?
Wesley Fink - Thursday, August 24, 2006 - link
The news is on the Gigabyte website at http://www.gigabyte.com.tw/News/Company/News_List....">http://www.gigabyte.com.tw/News/Company/News_List..... You find the Asus news on the Asus site at http://www.asus.com/news_show.aspx?id=3900">http://www.asus.com/news_show.aspx?id=3900.The "merger" was announced a few weeks ago in the Taiwan technical trade press.
xsilver - Friday, August 25, 2006 - link
oh, not exactly a "merger"more like a partnership
they will be producing a new set of boards together ala asrock type boards?
any news if this partnership will be producing budget boards or enthusiast type boards, the press release doesnt mention that.
has a name been decided yet?
giga-sus ? lol
as-byte? double lol
I suspect this will be more a asrock type venture, where cheap mass produced boards can be made without affecting the parent brand name if questions of quality arise
the benefits though as with asrock, have been the company being able to be more adventurous on designs and configurations?
JaredExtreme - Thursday, August 24, 2006 - link
When I first saw the title of this article I was extremely excited.I was thinking to myself: "Asus adds unlocking? Awesome! I'll finally be able unlock those multipliers upwards on some cheap Conroes!" I have a feeling a lot of other people had the same initial enthusiasm when they saw the title.Of course then I got to the point where it's mentioned that none of the Conroes unlocked upwards. Given that the technology mentioned in this article has existed for some time, I'm not sure why the article was even put on Anandtech.
AdamK47 3DS - Thursday, August 24, 2006 - link
I believe Anandtech needs to edit the article to aknowledge that Gigabyte was the first with the feature.Madellga - Thursday, August 24, 2006 - link
Wesley, you probably know this (at least Gary should know it, as he has threads on the CPU Forum).This is not an ASUS feature - at least on the P965 boards. Gigabyte was there before.
I have a DQ6 and the shipping BIOS F2 had it already.
With so many people on the forum using Gigabyte (including Gary), I wonder why do we have an article based on the Asus.....
ZachSaw - Thursday, August 24, 2006 - link
Exactly. The whole article has been singing in ASUS tune when in fact they've just copied someone else's feature! Shame shame.It's such an old feature to exploit EIST for multiplier adjustment that Gigabyte didn't even talked about it any more. Perhaps Wesley is excited ASUS has finally implemented that feature? :) ASUS fan-boy.
Gary Key - Thursday, August 24, 2006 - link
This is my fault. I had a miscommunication with Wes about the unlock features on the Gigabyte boards. Asus had figured out a way to go up and down which (we are trying to find the proper lot number to test) Gigabyte and others had not at this time. In my conversations with Wes I relayed the wrong statement about Gigabyte's capability in this area or my words were confusing about the EIST exploit. Asus will also implement this capability on the 975X chipset in the near future. If the tables were turned on the manufacturer then Wes would be a Gigabyte Fan-Boy I guess......
ZachSaw - Thursday, August 24, 2006 - link
In the meantime, do take the article offline.Find a bigger sample -- determine which ones can be "unlocked" upwards, which ones can't. Compare with Retail B2. Those are the ones most people are using and are available off the shelves. Otherwise, the article is completely meaningless as it is right now. Other than to make Wesley appear like an ASUS fan-boy, of course! ;)
Anandtech articles are so significant to a lot of people and companies -- you simply can't afford to make such elementary mistakes! If some motherboard manufacturers tell you they have unlocked one of Intel's most highly kept secrets, you'll have to start questioning from every angle -- not embrace it like every other fan-boys.
Like I've said, there're only 2 tools that are capable of 'unlocking' (it's actually reversing) the fused settings. 3 if you consider micro surgery on the chip itself. Never had the need to use that big expensive machine though. If the chip is properly configured (i.e. all production parts), you can forget reversing it. I can't go into more details about the 2 tools, but I can assure you that ASUS or any other mobo manufacturers for that matter do not have access to them.
Gary Key - Thursday, August 24, 2006 - link
We agree. :) We do have screenshots from Asus along with other engineering information showing chips upclocking. We are discussing the upclock statements but for the mean time the article has clarified the down clock information. That error was my fault as my communication to Wes was flawed when discussing the down clock with and without EIST enabled.
As for the B1 / B2 steppings, B1 was the planned retail release and several shipments did make it into distribution. Our B2 E6300/E6400 overclock better than their B1 counterparts. As for the B1 E6600~X6800, our internal units vary up to 200MHZ in overclock ability on the X6800 as an example. Intel sent out several thousand B1 review samples and that is what we have to use unless individuals on the staff purchase new retail chips which can become quite expensive. ;-) I have been through all of the overclocking results with our press samples/retail chips. Unless you had access to hand binned chips, then it truly is the luck of the draw about the chip capability you will receive.
Anemone - Sunday, August 27, 2006 - link
This is correct. B1's did make it to retail.Zach, for someone who's not supposed to say much, you talk a lot... and not by email, which I find surprising. It's your call, but it is very unusual, and motivated by something I think.
These are curious posts for sure.
Frumious1 - Thursday, August 24, 2006 - link
They never said they unlocked up. And GIVE THE B2 vs. B1 SHIT A REST!Did they say their B1 unlocked up? NO! Did they say their B2 unlocked up? NO! So what exactly does B2 give you over B1, a display in CPU-Z that saying B2 and probably a processor that you didn't have to pay for, that's what. ASUS says some chips do unlock up; I certainly don't expect AnandTech to run around purchasing chips to try and find out which ones will work.
So Gigabyte and ASUS unlock down, other boards use EIST to automatically drop to 6X (and only 6X as far as I'm aware) for power saving. BIOS updates can fix that, but a lot of companies won't bother. Anyway, at least we now know that you work for Intel, what with your L33T Inside Information.
ZachSaw - Thursday, August 24, 2006 - link
Read the post I responded to, kid, before you speak.Frumious1 - Thursday, August 24, 2006 - link
Thanks, old Intel fart who probably works in Malaysia for Intel at a chip fab. As I said, never say never. If there's one thing I can say for certain, it's that the Taiwanese, Chinese, etc. have some very talented engineers. I don't doubt that Asus has some non-ES, non-X6800 chips that unlock up. Are they simply an analomy? Possibly - probably even - but you "2 sooper secrit toolz" that allow reversal of the locks simply mean that there is a way to pull this off. Whether it's practical or not, I can't say, and you're not telling.Anyway, they've updated the conclusion to now mention Gigabyte, so I hope that keeps you happy. I for one must also be an Asus fanboy because I find their Intel boards to be among the best around - far better than the BadAxe, that's for sure. Without knowing anything about a motherboard, I'd go with chance and take an Asus over any other manufacturer. There have been a few lemons, but everyone has those - even Intel (and AMD, Gigabyte, etc.) By and large, I trust Asus more than any other motherboard company.
And hey, if they want to market their boards better than Gigabyte and specifically ask Anandtech to do a special report on the feature, more power to 'em! AMD could learn a thing or two about marketing from Asus. Heh. I still prefer the P5W DH, but would really like to see SLI support with a high-end chipset. (Yeah, I'm an nVidia fanboy as well - not my fault that ATI's drivers are still flaky in a lot of situations and have a cruddy UI.) Unfortunately, I don't think 590 SLI is going to be the answer... maybe "nforce6" or whatever will offer high overclocks and SLI support, but that's still a while off it seems.
ZachSaw - Thursday, August 24, 2006 - link
Nope. I don't work for Intel.And no, you're way off the point -- this is never about Gigabyte vs ASUS.
And yes, you're entitled to have your own opinion. Perhaps one day we'll even have a mobo that allows you to have both Intel and AMD chips on one board doing dual processor. Great, now off you go.
Frumious1 - Thursday, August 24, 2006 - link
Okay, so you don't work for Intel but you apparently have access to a microsurgery chip corrector machine, and you have access to tools that "ASUS or any other mobo manufacturers" lack. Wow. If you don't work for Intel, then I would have to conclude you either work for AMD, IBM, or some other major chip fabrication company (TSMC maybe). Then we of course have to wonder why this upsets you so... but if you work for AMD that would be a given, since good news for Intel is bad news for AMD generally speaking. I have difficulty imagining something that you might have access to that Asus would not.
I suppose you need to protect your identity, regardless, but it's all rather silly to bash a short article that basically says "Asus claims upward unlocks are sometimes possible" when you really don't have proof, except from your vast knowledge of what is and isn't possible. The world is full of things that were once deemed "impossible", and if Anandtech says ASUS has sent them screencaptures, I have no reason to doubt that. Were they with ES, B1, B2, or something else? Nobody has yet given a concrete answer.
Do I trust you, or Anandtech? I forgot, AT is full of Intel, AMD, Asus, Gigabyte, ASRock, etc. fanboys, so I guess we should all trust Zach instead. Great, now off you go. Hope they pay you well. Goodnight.
ZachSaw - Thursday, August 24, 2006 - link
There's no way you can unlock it upwards -- unless you have the 2 tools I had access to. And no, you do not have access to those tools.The only way ASUS managed to 'unlock' is simply because Intel did not fuse out the multiplier in certain engineering samples. There're different classes of samples. I am not allowed to go into too much details, but the article is simply wrong.
Anyway, if he sang so much praises for Gigabyte in one article, yeah, I'll call him a Gigabyte fan-boy. :) Since he can't proved that ASUS unlocked it upwards on Retail processors, he ignorantly praised ASUS nevertheless. You'd have to agree that it warrants the label of fan-boy! ;)
Frumious1 - Thursday, August 24, 2006 - link
Only thing I can say for certain is that you're coming off as an arrogant prick, Zach. I sure hope you work for a motherboard/RAM/CPU manufacturer. As for what is and isn't possible with BIOS tweaks, never say never. Maybe Intel just had a batch that didn't get locked properly, maybe not.But a FSB overclock to 532 MHz is definitely nothing to scoff at, and whether Gigabyte was first with a buggy BIOS or not doesn't matter much to me. After all, the ftp://dlsvr01.asus.com/pub/ASUS/mb/socket775/P5B%2...">0507 Asus BIOS is from 2006/8/11 which makes it two weeks old. That's about the time Core 2 really became available anyway. Gigabyte's F4 came out around the same time, and that's the first truly usable BIOS from them as far as I can tell.
ZachSaw - Thursday, August 24, 2006 - link
If you're in my position, knowing so much without the liberty of sharing the knowledge, you'll know how I feel. So it's not arrogance. I simply am not allowed to disclose too much. And do feel free to continue speculating. I've allowed you guys to go on for years without intervention, but this simply crosses the line. Anyhow, you probably won't be hearing much from me any more. If I've appeared to you as an arrogant prick trying to guide the authors to the correct path, then so be it.
atenza - Friday, August 25, 2006 - link
I agree with you. I'm sure there is absolutely no way to change the multiplier UPWARDS on a stock E6300-E6700. Given the overclocking potential of any of these CPUs, everyone could simply make his cheaper CPU a X6800 (or X6800 with 2MB cache :) And this really is the last thing Intel wants.Anemone - Thursday, August 24, 2006 - link
The article wasn't really so bad as the criticisms seem to tell, imo. Abusing IST has been around a while I admit but the basis of a stronger memory controller bios still stands and this board is significantly stronger than Gigabyte's in that regard alone.There were B1 steppings that made it to retail, at least according to a few of the recording threads I've seen.
I think the only thing truly sad is that the bios's have been abhorrently buggy on the 975's the 965's and the 590's. People have started to measure which is the "least buggy" which is kind of crazy at the prices these boards are commanding. People need to let these things cool a bit, and mature a LOT, before they're going to be worth what's being asked for them. Only after things have matured and are tested are we really going to see which of all is worthy of the money. And sadly, from what I'm hearing, the 590 is definitely NOT going to be it, so better keep an eye on the other two.
ZachSaw - Thursday, August 24, 2006 - link
Nope.
It's true that Gigabyte sucks when it comes to memory compatibility. But this article is not about that, is it?
If Gigabyte fixes the issues on memory compatibility, and the author writes about it saying that it's the first manufacturer to come out with a bios that is highly compatible with a lot of memory modules, I'll still be ranting about it.
People should own up to their mistakes.
ZachSaw - Thursday, August 24, 2006 - link
Well, perhaps the author shouldn't jump at the very first chance to suck the toe of ASUS' BIOS programmer... check around forums to find out if other manufacturers have this feature as well. A little research is always required before you start writing an article that will be read by thousands of people. It's the author's obligation.Madellga - Thursday, August 24, 2006 - link
Specially when Gary knows that. A coffee chat would do it.bob661 - Thursday, August 24, 2006 - link
Give me SLI on this chipset and I'll kick the upcoming 590 to the curb!Atty - Thursday, August 24, 2006 - link
If ASUS has added this feature for the P5B does that mean that the other boards for Core 2 (most importantly for me the nForce 500's) will have this feature?ZachSaw - Thursday, August 24, 2006 - link
Why's everyone so worried about other boards not having it?It's a very simple addition. Gimme the BIOS source, I'll add it in for u. It's a default 'feature' of the CPU, so whether or not you get it in your board will depend on how diligent your mobo's BIOS programmer is. Shouldn't take longer than 30 minutes to implement and test (internally of course). It's the process of complete validation of a new release that most BIOS programmers fear.
mpc7488 - Thursday, August 24, 2006 - link
Great results, very exciting.ZachSaw - Thursday, August 24, 2006 - link
"the ability to unlock the previously locked Conroe chips is so significant that every other manufacturer will have to follow the ASUS lead or lose market share to the motherboard giant."LOL. Perhaps ASUS is the one who is the follower. All of Gigabyte's 965 boards have had this ability since the very beginning. ASUS had to release an update to their BIOS to support this, after seeing Gigabyte's feature.
In any event, this is a VERY SIMPLE function to implement -- what innovation? It simply uses the EIST feature in your CPU to implement multiplier "unlock". In fact, it's technically wrong to say "unlock", since we don't lock the multiplier you managed to select in the first place! That's the whole point of EIST. Unless you can select upwards, then you can claim to have unlocked the multiplier. Otherwise, it's just legally and technically a wrong term to use. You can claim to support Multiplier Adjustment, but do not EVER say it's unlocked. Be warned.
Please clean up your article. Too many mistakes.
Also, it'll be good to put a foreword in your article to mention that your oc is based on Engineering Samples (B1 step), which are different from retail (B2). Retail ones have some differences which make it a little faster clock to clock, but not as overclockable. You do not want people to rush out to buy those processors, based on your table of overclocked frequencies.
Madellga - Thursday, August 24, 2006 - link
Correct. Reviews like this should be based on Retail samples, not Golden Cherry Picked units.Otherwise people are misled by the results.
Frumious1 - Thursday, August 24, 2006 - link
Yeah, "Golden Cherry Picked units" like the freaking RETAIL E6400 and E6300 stepping B2 that overclocked higher than the "cherry" ES B1 E6600/E6700? OMFG the world is coming to an end!!!The only thing potentially misleading is that the B2 E66/6700 might do better than shown here. The bottom line is that the B2 is a later, IMPROVED stepping that should actually at least match these results.
The Gigabyte information is more of a problem, but given my experience with a DS3 I can't say I'm too happy with GB right now. F4 seems to have cleared up most of the problems. F2 was an absolute joke, however.
ZachSaw - Thursday, August 24, 2006 - link
You simply do not know enough. B2 is a later step, no doubt. But, who says later steps increase oc-potential??? It's pretty weird how the community came to this "conclusion". It might be true for later steps (C and above), but usually not the earlier ones such as B1 and B2. So don't jump to conclusions when you have never tested out A0 parts before.
All I'm asking them to do is, to use retail parts. That's all.
Frumious1 - Thursday, August 24, 2006 - link
Maybe you can pull some strings with Intel and get them to send new B2 or later retail samples to Anandtech? I'm sure the AT guys don't want to go out and buy a bunch of new CPUs when the current ones work fine. We all recognize that there is an element of marketing to hardware enthusiast sites - that's why there are a shitload of ads at all the sites, right? (Thank you, AdBlock!) AMD and Intel (and Asus, Gigabyte, Asrock, etc.) send parts for free to AT and they review them. "No such thing as bad publicity!"You keep ignoring the fact that the E63/6400 were NOT B1/ES steppings, and yet they overclocked better. That's where I drew my conclusions. Sure, less cache has an impact, but at the very least it's safe to say B1 isn't amazingly better than B2. Anyway, all overclocks are largely CPU dependent. Everyone in the community knows that. YMMV: Your Mileage May Vary. That the P5B hits 500+ MHz FSB is more than enough for me. I don't particularly like to run at "mostly stable" settings, so if I can POST and run most benches at a 68% OC, I'll drop to 50% for 100% stability and peace of mind. $185-$200 for a chip that can match stock X6800 performance is pretty tasty.
shecknoscopy - Thursday, August 24, 2006 - link
1) How is it possible that they can do this? I thought the multiplier lock was something phyiscally hard-wired in the manufacture of the processor - akin to those graphics cards that absolutely cannot unlock extra pipes.2) Any chance that we'll see it appearing on the other ASUS conroe boards? Or is the answer to (1) intricately linked to the north bridge?
3) How about other manufacturers? Now that ASUS has set the stage, are we going to see the Gigabyte (or what's left of 'em), MSI, DFI, etc... boards following suit? Or is this patentable technology?
4) Anyone else had it with these <explicative> snakes on this <explicative> plane? I wish someone would do something about them, already....
-Sheq
ZachSaw - Thursday, August 24, 2006 - link
Write a few bytes into the MSR. Read the PRM of conroe. You'll find out how easy it actually is. Under Windows, you'll need a driver's help to access MSRs, but the BIOS has complete access to MSRs.
ASUS followed Gigabyte -- AFAIK, Gigabyte already has this feature since they introduced 965P-series boards. They did it so early that they didn't need any updates.
Sh0ckwave - Thursday, August 24, 2006 - link
I can confirm this. I just tried it on my DS3 with and E6400. I was able to set the multi to 6, 7, and 8.Calin - Friday, August 25, 2006 - link
Who knows, after the Gigabyte-Asus union (partial union with a manufacturing purpose), maybe they do share some know-howtombomba2 - Thursday, August 24, 2006 - link
Sorry, but as long as anandtech doesn't have conroes which can be unlocked up, I think this is a very questionable feature of the board.Wesley Fink - Thursday, August 24, 2006 - link
We agree with you. We stated what we found - all Conroes we tested unlocked down. We also told you what Asus told us, but we were clear it could not be confirmed in our own testing. We did see screen shots of a locked Conroe at higher multipliers, but we did not find one in our testing. Our coverage was fair IMHO.Sunrise089 - Friday, August 25, 2006 - link
I would say the coverage was helpful, but not fair. Or if it was fair, it wasn't what I except from AT. I think normally you guys are pretty tough on campanies that don't deliver on promises, which is great because:-You're an advocate for the consumer
-You serve as a counterweight to huge manufacturer-funded marketing efforts.
I think your position on promised video-card launch dates that came and went was more what I like - you reported the manufacturer's stated date, but also blasted the GPU maker for letting the dates slip in the past and suggested it was likely to happen again.
It's fine to publish ASUS's claims, after all you want them to send you products to review, but why not editorialize a little more. Not have one chip that did what they claimed was sure plenty of justification.
Sapiens - Saturday, August 26, 2006 - link
So we should blame ASUS for the lack of Conroe chips at AnandTech that can be unlocked upwards? AT stated ASUS' claims and warned that they weren't able to verify all of them. It sounds like you just want the authors of the article to bash someone.Sunrise089 - Saturday, August 26, 2006 - link
I want the authors to keep up working on behalf of the consumer. When a company touts a feature that isn't present in many if not most cases, the company shouldn't get much credit for the "feature". Repeating the party line offered by ASUS gives them more credibility than they need to be given here. You and I can all go to ASUS's site and read up on their marketing department's list of reasons to buy their board, I expect AT to revivew the board with an eye to using their expertiese to inform me about it's strengths and weaknesses. An advertised feature that is present in ZERO percent of tested boards is more of a minus than a plus.