2.8 LGA775 Retail CPU Overclock

FrgMstr

FrgMstr

Just Plain Mean
Staff member
Joined
May 18, 1997
Messages
55,601
(UPDATE: The full article has been posted on HardOCP as of this morning. Please feel free to discuss our testing and results.)


Just wanted to give you guys a heads up. NewEgg was kind enough to send me a 2.8GHz LGA775 CPU for OC testing. If you check our ABIT AA8 (i925X) review, you will see that 240FSB was a stretch for our engineering sample CPUs. The retail 2.8 did a 250MHz FSB at default settings with Corsair DDR2. With less than .1v bump to the Vcore, we have been running at a SOLID 255MHz x 14 for a full day now.

Anyways, a 1018MHz FSB and 680MHz memory bus are sort of nice chugging along at 3.5GHz. :)
 
  • Like
Reactions: HAF72
like this
nice. wonder wut type of scores that system would get in some mem benches. running at 4-4-4-11?
 
I am working through some PCI-Express issues right now with particular cards. Will get to that in a bit.
 
Thanks for testing, its interesting that you could reach 255MHz FSB, not so much that the CPU can do it, but rather the motherboard, as it seems a lot of people so far have hit the 248MHz FSB wall (as did Keith originally). Did the motherboard require any increase in NB voltage to hit 255MHz?
 
ehhh, i saw overclocks better than that on northwoods a year ago. i dont see what the hype is.
 
ya but where are the pcie 6800 gts from newegg to go in my soon to be aa8? :(
 
jamestime88 said:
ehhh, i saw overclocks better than that on northwoods a year ago. i dont see what the hype is.
Click to expand...
No hype here, simply reporting the facts. I think this can hardly be seen a "bad news"....
 
Is it possible that there will be revisions of the PCI-E GeForce GT cards that will relax the power consumption requirements and lessen the overclocking issues you encountered?

The AGP 6800 GT seems more flexible and able to run with lower rated power supplies.

I'm specifically interested in the SFF angle. Many people seem to be running 6800 GTs in 200-250W SFF cases and even overclocking both CPU and GPU without problems.

From reading about your issues, it seems like a PCI-E 6800 GT would be out of the question for a 915/925 SFF.

Any thoughts, or will the PCI-E X800 be the only real choice for a SFF?
 
You mentioned having an Asus LGA775 board in this article, are you planning on releasing a review of this board soon?

berzerker said:
Is it possible that there will be revisions of the PCI-E GeForce GT cards that will relax the power consumption requirements and lessen the overclocking issues you encountered?

The AGP 6800 GT seems more flexible and able to run with lower rated power supplies.

I'm specifically interested in the SFF angle. Many people seem to be running 6800 GTs in 200-250W SFF cases and even overclocking both CPU and GPU without problems.

From reading about your issues, it seems like a PCI-E 6800 GT would be out of the question for a 915/925 SFF.

Any thoughts, or will the PCI-E X800 be the only real choice for a SFF?
Click to expand...

VGPU can be reduced through modded BIOS (either by OEM or User), which would allow for lesser power draw. Of course, reduced VGPU may affect stability at high GPU clocks, and one should know what they are doing whenever they reflash any BIOS.
 
berzerker said:
Is it possible that there will be revisions of the PCI-E GeForce GT cards that will relax the power consumption requirements and lessen the overclocking issues you encountered?

The AGP 6800 GT seems more flexible and able to run with lower rated power supplies.

I'm specifically interested in the SFF angle. Many people seem to be running 6800 GTs in 200-250W SFF cases and even overclocking both CPU and GPU without problems.

From reading about your issues, it seems like a PCI-E 6800 GT would be out of the question for a 915/925 SFF.

Any thoughts, or will the PCI-E X800 be the only real choice for a SFF?
Click to expand...
Right now I really don't know enough about it or have enough experience to even guess right. I think a lot of the issues here are going to come down to motherboard quality, but running a NV45 in a SFF box with a 250w PSU seems to be iffy under a lot of stress.
 
Danny_Ocean said:
You mentioned having an Asus LGA775 board in this article, are you planning on releasing a review of this board soon?
Click to expand...
Yes, will likely be next week. We still have some PCI-Express stability testing to do on it as well.
 
Kyle, since it seems that a quality PS is going to be much more of a factor in the future, are you planning to start doing PS reviews?
You almost did a PS mini-review inside this review. ;)

Seems like people are always willing to spend $$$ on the latest / hottest video card, mobo or CPU but balk at spending $100-$200 on a quality PS.
I've never understood that mentality. :rolleyes:
 
qdemn7 said:
Kyle, since it seems that a quality PS is going to be much more of a factor in the future, are you planning to start doing PS reviews?
You almost did a PS mini-review inside this review. ;)

Seems like people are always willing to spend $$$ on the latest / hottest video card, mobo or CPU but balk at spending $100-$200 on a quality PS.
I've never understood that mentality. :rolleyes:
Click to expand...

When u spent 200 dollars on a video card, you instantly see a performance increase; however, with a power supply, what do you really get?
 
mjz_5 said:
When u spent 200 dollars on a video card, you instantly see a performance increase; however, with a power supply, what do you really get?
Click to expand...
You get a system that you can O/C to the limits of your CPU / mobo / memory or your video card. You get something that will last you quite a while. A good PS is an investment that can serve you for years, it'll be around long after you've replaced everything else, with the exception of maybe your case. You don't see people saying "I just spent $200 on a PC Power and Cooling PS last year, now I need a new PS", do you? You see people saying just that with CPUs, video cards, or motherboards.

It would be ignorant to spend $500 on a brand new top-end video card that you can't take advantage of because of a POS PS, yet people do just that.
People drop $$$$ into systems, then try to make do with the PS that came with their $99 case, and then post that they can't O/C their system.

Anything in life is only as strong as it's weakest link. This is true of computers as well as anything. If your PS is crap, then I don't care what else you've got, you're system is crap. The PS is without a doubt the most ignored component in the average enthusiast system, IMO.

Ask yourself this; Are you willing to risk $2000 worth of CPU / mobo / video card / memory and drives on a crap PS?
 
qdemn7 said:
You get a system that you can O/C to the limits of your CPU / mobo / memory or your video card. You get something that will last you quite a while. A good PS is an investment that can serve you for years, it'll be around long after you've replaced everything else, with the exception of maybe your case. You don't see people saying "I just spent $200 on a PC Power and Cooling PS last year, now I need a new PS", do you? You see people saying just that with CPUs, video cards, or motherboards.

It would be ignorant to spend $500 on a brand new top-end video card that you can't take advantage of because of a POS PS, yet people do just that.
People drop $$$$ into systems, then try to make do with the PS that came with their $99 case, and then post that they can't O/C their system.

Anything in life is only as strong as it's weakest link. This is true of computers as well as anything. If your PS is crap, then I don't care what else you've got, you're system is crap. The PS is without a doubt the most ignored component in the average enthusiast system, IMO.

Ask yourself this; Are you willing to risk $2000 worth of CPU / mobo / video card / memory and drives on a crap PS?
Click to expand...

I know what you are saying, I went through 3 systems and I still have my trusty Lian Li case and Enermax 430 Watt power supply

Another example is a monitor, some people spent 500 dollars on a video card and continue using their 14" CRT monitor they had when win 3.1 came out. Well, that is changing now, but that how I use to remember it.
 
Our 2.8GHz part is locked at "14". 14 X 2800MHz = 2.8GHz.
Click to expand...

Don't you mean 14 * 200MHz = 2.8GHz Kyle?

Great review though, Im drooling over all this new hardware coming out that has strayed so far from what hs been in place for years
 
Awesome review Kyle, and the timing couldn't be better. I ordered the same hardware you used for the review last week with the exception of a 3.4GHz Prescott and an Antec TRUE550. I'm interested to know which 6800 Ultra brand video card was used. At the time of this writing, the only retail 6800 Ultra that clearly advertises support for PCI-e is MSI's NX6800 Ultra, which I have preordered. Are all 6800 Ultras using Nvidia's bridge chip and it's just not advertised, or do you have an actual NV45 sample? I can't find the MSI in stock anywhere, so I had to preorder it and wait until the 23rd for it to ship. Can't wait to do some OC!
 
I just purchased a 2.8e along with a gigabyte 915p duo-A. Hopefully I'll have time tonight to mess around with it. I do not, however, have any sort of watercooling whatsoever. I'll also see how well the GEAR agp slot works.

-Steve
 
qdemn7 said:
Kyle, since it seems that a quality PS is going to be much more of a factor in the future, are you planning to start doing PS reviews?
You almost did a PS mini-review inside this review. ;)

Seems like people are always willing to spend $$$ on the latest / hottest video card, mobo or CPU but balk at spending $100-$200 on a quality PS.
I've never understood that mentality. :rolleyes:
Click to expand...
You know, we have kicked around doing PSU reviews for a long time now, but the fact of the matter is that I think you need at least 3 months of testing to REALLY see what a PSU is all about. Longevity if the key factor in buying a good PSU and that is something that is very hard to test.
 
Biff said:
Awesome review Kyle, and the timing couldn't be better. I ordered the same hardware you used for the review last week with the exception of a 3.4GHz Prescott and an Antec TRUE550. I'm interested to know which 6800 Ultra brand video card was used. At the time of this writing, the only retail 6800 Ultra that clearly advertises support for PCI-e is MSI's NX6800 Ultra, which I have preordered. Are all 6800 Ultras using Nvidia's bridge chip and it's just not advertised, or do you have an actual NV45 sample? I can't find the MSI in stock anywhere, so I had to preorder it and wait until the 23rd for it to ship. Can't wait to do some OC!
Click to expand...
The card was directly supplied by NVIDIA.

All NV45 use a bridge chip. The core is a native AGP part.
 
Yet another excellent review, Kyle. Thanks for the info. Looks like the 1066FSB might actually be worth waiting for and having.... As long as Intel doesn't hobble it like they tried to do with this generation of chipset.
 
gimper48 said:
http://www.anandtech.com/cpuchipsets/showdoc.aspx?i=2124

This gives some interesting insite to the overclock lock.
Click to expand...
And there are some things in there as well that I think are just wrong. I think his #2 statement is incorrect and so is #3. I have proven different here on our test equipment. We are OCing far beyond his 10% with no tweaks at all with a retail CPU. We saw the same issues with the engineering sample CPU.

The SATA does have a lot to do with it though. And I do agree with his statement that ATI seems to be more tolerant, at least on the newer VPUs.

It looks as though one thing is for sure, solid OC boards from "everyone" is not going to happen on the 925/915 front. With the latest Asus BIOS, our OCs actually got worse not better. There is still a lot of half-truths everywhere on this. That is one reason we did not get into theorizing explanations that we were not sure of backed up with limited test data using just engineering sample CPUs.
 
ss284 said:
I just purchased a 2.8e along with a gigabyte 915p duo-A. Hopefully I'll have time tonight to mess around with it. I do not, however, have any sort of watercooling whatsoever. I'll also see how well the GEAR agp slot works.

-Steve
Click to expand...
Gigabyte will not send us a board to test yet, that usually means there are going to be OC issues with it. PLEASE let us know what you find out.
 
Kyle,
I would also be very interested in power supply reviews - With all the manufacturer's claims and the need for rock solid voltage levels under ever increasing loads, the PSU is becoming the "bottleneck" much more often. Also Power supplies have almost become a defacto "peripheral" - Builders choose the PSU independant of case to meet specific requirements, it would be nice to seperate the wheat from the chaff, as it were.
 
qdemn7 said:
Well what do you think of Silent PC Review's method of load testing? I know that noise may not be as much of a factor here as there. But otherwise IMO it looks like a good method.

I'm asking because I don't realistically see how you could do a 3 month test without driving yourself nuts, or costing $10K worth of systems set up and running flat out until the PSU craps out and frys $XK amout of hardware.

There has got to be a reliable, reproducable method of testing PSUs without taking 3 months to do so.
Click to expand...
They have access to some very nice equipment for sure, but I am unsure that their method looks really well at PSU longevity and that is what it is all about beyond the obvious needs of doing what it is speced to do.

And again, we have not done PSU testing because it is not economical or realistic to do it how it is NEEDED to do.

I think the best PSU test is word of mouth. I have recently moved from Enermax to Antec PSUs. We have several servers with Antec PSUs in there that have been solid for well over a year now and a 550w PSU pulling two CPUs, 4GB of ram, and 9 15K RPM SCIS hard drives and all the motherboard has to be tough. :)
 
one thing i think you didnt mention was the price of ddr2- its more than double the price of current ddr. im not sure if its so easy to recommend this setup to enthusiasts, as a new mobo, cpu, and ram will cost around 800 dollars
 
wtf said:
one thing i think you didnt mention was the price of ddr2- its more than double the price of current ddr. im not sure if its so easy to recommend this setup to enthusiasts, as a new mobo, cpu, and ram will cost around 800 dollars
Click to expand...

I agree. Right now there is no compelling reason IMO to jump to socket 775. In my opinion users would be much better off staying with AGP cards, DDR PC3200/3500/etc... and preferrably an A64 or P4 Northwood setup depending on user preference and primary uses. To me the Prescott is a failure at anything below 4ghz, and at that point it becomes ok, approaching near Northwood performance clock for clock. If a user has a POS year 2000 dell, then if they want to spend $800 on a socket 775 setup I guess that would work, or they could spend $800 on a nice socket 939 3500+ setup oc it to 2.5-2.6ghz on air and whallop that prescott into submission. Or they could get a P4C at 2.8-3.0ghz, oc it to 3.6ghz farily easy and a nice 875 board run about 10-15C cooler, use cheaper DDR ram and still experience faster gaming.

I realize you(at HardOCP) are excited about 775, as I'm sure many are. My question is why there are comments in conclusion that this is such a viable alternative whereas it seems invariable that any A64 review is met with the conclusion that while it's fast it's not worth the $$$ and a oc'd 2.4C seems to garner your recommendatoin.

Just confused on this. Are you seeing tangible difference in 775 to warrant added costs over a cheaper P4C or A64 in 754 or 939 setup?????
 
Wow 150 degrees is pretty hot :eek:

Let's see this thing coupled with some phase change cooling! :D :cool:
 
wtf said:
one thing i think you didnt mention was the price of ddr2- its more than double the price of current ddr. im not sure if its so easy to recommend this setup to enthusiasts, as a new mobo, cpu, and ram will cost around 800 dollars
Click to expand...

Kingston Value-RAM DDR2-533 costs $333 for a 1GB Kit (2x512MB) from http://shop.kingston.com/save/.

Crucial DDR2-533 sells for $159/512MB at DLJ System and $187/512MB at www.Crucial.com.

These are reasonable prices for products in limited supply elsewhere and which have just entered the market. High-cost DDR from the likes of OCZ and Corsair sell for above $340.
 
alright, even if the ram is same price, there is still no tangible performance advantage of running ddr 2 IMO. Maybe sandra could post higher numbers but overall still not enough to make up for poor performing prescott in relation to northwood. Plus factor in that most people jumping on these performance boards probably already have a good agp card, good ddr ram it's even more perplexing to me why they would do that(move to slower prescott, move to PCI express which shows no current advantage, and DDR-2 which at least isn't slower than DDR. Course the same argument could be made for me going socket 939 and FX53, but despite the high cost it's at least faster than older socket design :p
 
Danny_Ocean said:
Kingston Value-RAM DDR2-533 costs $333 for a 1GB Kit (2x512MB) from http://shop.kingston.com/save/.

Crucial DDR2-533 sells for $159/512MB at DLJ System and $187/512MB at www.Crucial.com.

These are reasonable prices for products in limited supply elsewhere and which have just entered the market. High-cost DDR from the likes of OCZ and Corsair sell for above $340.
Click to expand...

thats ddr2 533 though, the 667mhz stuff used in hardocp's tests costs much more
 
RagingSamster said:
Kyle,
I would also be very interested in power supply reviews - With all the manufacturer's claims and the need for rock solid voltage levels under ever increasing loads, the PSU is becoming the "bottleneck" much more often. Also Power supplies have almost become a defacto "peripheral" - Builders choose the PSU independant of case to meet specific requirements, it would be nice to seperate the wheat from the chaff, as it were.
Click to expand...


http://www.hardforum.com/forumdisplay.php?f=93
 
wtf said:
thats ddr2 533 though, the 667mhz stuff used in hardocp's tests costs much more
Click to expand...

Corsair DDR2-667 should be using Micron's 3.7ns chips, which are the same found on Crucial DDR2-533. So you get the same performance minus the pretty lights. These, and most other DDR2 chips, are showing themselves as great yielders.
 
You must log in or register to reply here.
Back
Top