DFI 875P-T @ 6Ghz

Printable View

Show 100 post(s) from this thread on one page

11-12-2004, 04:55 PM
meke

Quote:

Originally Posted by IntelInside

Here you go:

Prescott PI
Northwood PI
Technical details of the patch

Everyone, sorry I got a bit pissed off. Fugger, great job done with O/C. Keep it up.

I patched the super_pi with the prescott_pi but unfortunately I got error and cannot continue.

is this normal?

http://img.photobucket.com/albums/v2...erPI/error.jpg
11-12-2004, 04:58 PM
Charles Wirth

No LN2. I picked up a new X800XT and I will put the PE bios on it.

Got lucky af Frys, they had the Visiontek X800XT in stock and it has 1.6ns ram and Rage Theatre chip on it.

Ill try not to crush the core this time ;)

My 3.8E E0 ES will be here monday but I will still get in some runs on the 3.6E E0 in the mean time.

Adding a 600w OCZ PSU to the mix

3.6E E0
DFI 875P-T
Visiontek X800XT modded
OCZ Powerstream 600w
OCZ Rev2 TCCD
OCZ DDR Booster

And just for the fun of it all I may abuse a BFG6800U as well.
11-12-2004, 04:59 PM
Charles Wirth

Meke if you are overclocked too far it will caugh up that error message.
11-12-2004, 05:30 PM
pik-ard v1.1

fugger i've heard bad things about the 600W OCZ, i've seen reports of a few poping... might wanna wait with that untill OCZ gets whatever problem it has worked out.

just a heads up.
11-12-2004, 06:11 PM
IntelInside

Impatient people should skip this post because it is going to be long...

Quote:

Originally Posted by chilly1

And in the original superPI is there any other code that is processor unique?

Original SuperPI is compiled for generic Pentium even without MMX.

Quote:

Originally Posted by Kane

IntelInside has a point. If we can't use his SuperPI, because it uses SSE3, why should people be allowed to use HT in benchmarks? What's legal, and what's not? Personally I don't se it as a cheat. Don't know about you guys, but would be fun to hear your opinions. Oh, by the way, I own a AMD, so I'm not a FanBoy

You said it better than me. I agree 101%.

Quote:

Originally Posted by FUGGER

Pi is a touchy issue for most when it comes to version and optimizations. we had a version we trusted the result but had very poor ability show exact results. meaning that it did not break down to tenths or hundredth of a second.

As far as I know original SuperPI also does not show 1/10 or 1/100 of a second. So what exactly is your point here?

Quote:

Originally Posted by Xassius

as i have mentioned before...do we need DOOM 3 Prescott Version and DOOM3 Winchester Version?

I will answer this with a question -- If it performs better that way then why the hell not?

Example: Tomb Raider -- Angel of Darkness has 3 EXE files. One generic, one for Pentium 3 and Athlon (SSE), and one for Pentium 4 and A64 (SSE2).

It is also possible to produce single executables that run optimized versions of critical code paths depending on the CPU you have.

Or lets make it even simpler for you -- if you pay for expensive HDTV capable set will you agree to only watch regular TV program on it?

Quote:

Originally Posted by STEvil

I have no problem with people running it, just as long as it can be distinguished from "normal" super-pi.

Problem is that SuperPI should have been written in such a way that it uses the maximum potential of each CPU. Much like VirtualDub for example. Soon reviewers that fancy AMD CPUs will start using VirtualDub as a benchmark tool because Avery Lee has ported it to 64 bits and those who will then say "it is not fair, they must use 32 bit version" will be laughed at. It is inevitable that the tests evolve along with the hardware or they become meaningless.

SuperPI for example only uses FPU unit (floating point calculations). It could have been much faster if it had optimizations for MMX, 3DNow, SSE, SSE2, SSE3, x86-64, etc. Everyone would then be able test their maximal CPU strength -- that is what we really should compare. That is not favoring one or the other, it is just that everyone is using what he paid for.

Imagine two friends comparing the power of their cars. One has the engine with 4 cyl. and another one has the engine with 8 cyl. Should he disable half of his engine for the comparison to be fair? I don't think so. After all they are not comparing the same two cars much like we do not compare the same CPUs here.

Quote:

Originally Posted by STEvil

maybe its time for a new version of super-pi?

I would suggest taking FastPI43 and building upon that.

Quote:

Originally Posted by STEvil

Myself I want to see a video card calculate pi to 1m places..

It is possible to do that and the library for doing things on the GPU called brook is under development so expect to see things like that pretty soon. For comparison Geforce 6800 Ultra can peak at 40 GFLOPS as compared with only up to 6 GFLOPS in theory on currently fastest CPUs.

Quote:

Originally Posted by CodeRed

This is going to make it very hard if not impossible to compare superpi scores, thus invalidating it as a benchmark.

benchmark(ing):
n., A standard by which something can be measured or judged
tr.v., To measure (a rival's product) according to specified standards in order to compare it with and improve one's own product.

Ok, lets see what makes up for a standard in a SuperPI case:

1. Calculation to a certain number of decimal places (usually 1 M)
2. Exact result
3. Time it took to complete.

If you are not satisfied with the above checklist then who will decide what additional rules should make the standard?

Someone could insist on other things such as:

1. Testing under Windows 98
2. Providing a list of running applications and services
3. Using same type and same amount of RAM
4. Using same chipset
5. Disabling Hyper-Threading if Intel
6. Using same amount of L1 and L2 cache
7. Starting the test after exactly x minutes of machine uptime after cold boot
8. ...

I exaggerated some points in the hope that you will see how stupid that would be.

As OPPAINTER said SuperPI sucks. Why? Because it only measures FPU strength which is almost irrelevant these days when most mainstream apps do not use FPU code but instead use optimized SSE and SSE2 code for calculations.

Quote:

Originally Posted by Xassius

excuse me...but why have you offically wrecked a benchmark many of us use?

That benchmark was meaningless anyway. See above.

Quote:

Originally Posted by Xassius

by no means am I saying that enabling optimizations in a program is bad...it's just that enabling them in a Benchmark for COMPARISON purposes and NOT BEING ABLE TO DIFFER BETWEEN PATCHED AND UNPATCHED is really wrecking and destroying the purpose of superpi...

Original purpose of SuperPI was to calculate PI to certain number of decimal points fast and accurate. It was not intended to be used as a benchmark anyway. Face it. It is skewed. AMD has faster FPU and that is why it performs better in SuperPI. That doesn't mean that in general use Pentium 4 is that much slower. Not all apps use FPU in that way.

Quote:

Originally Posted by meke

I patched the super_pi with the prescott_pi but unfortunately I got error and cannot continue.

is this normal?

As FUGGER says it is probably related to you O/C-ing the CPU too high. Try lowering the clock a bit.

Just for the record, my SuperPI result is 42s @ stock 2.8 GHz.
11-12-2004, 06:22 PM
pik-ard v1.1

IntelInside, i think the problem is, is that its only been optomized for ONE type of processor, so far as i know, theres no SSE2 optimization, MMX, 3DNOW, etc... so optomizing a program that is widely used in a community for just ONE of the contenders is unfair. what would have made people more agreeable to your modification, is if you got togeather with people who had other types of cpu's, and created a version of superpi with optimizations for ALL of them.

the point is that the optimization is only an optimization for a certain processor type, which renders it uncomparable to other processors.

maybe you think it is useless, but thats only your personal opinion, many others think that it is a good guage of brute strength of a cpu.
11-12-2004, 06:41 PM
chilly1

NO not for a while yet.... need a 200L dewer for what I want to do.
11-12-2004, 08:22 PM
CodeRed

Quote:

Originally Posted by IntelInside

...

Ok, lets see what makes up for a standard in a SuperPI case:

1. Calculation to a certain number of decimal places (usually 1 M)
2. Exact result
3. Time it took to complete.

If you are not satisfied with the above checklist then who will decide what additional rules should make the standard?

But that definition I could run pifast for 1M places and absolutely smoke any superpi score :rolleyes:
11-12-2004, 09:52 PM
Xassius

Quote:

Originally Posted by CodeRed

But that definition I could run pifast for 1M places and absolutely smoke any superpi score :rolleyes:

;)

:stick: @ intelinside

Intelinside: if it was a crap benchmark anyway...why did you patch it? It's only crap in your opinion....

people are mainly using superpi to compare the same group of cpu's...like Prescott -> another Prescott...so patching it would skew results...since you cannot distinguish between the two...
11-13-2004, 12:31 AM
meke

OK there is no change but I make it run this morning.

it's still 14x275 = 3850 Mhz

probably stock cooler was too hot yesterday :D

http://img.photobucket.com/albums/v2...b275_patch.jpg
11-13-2004, 12:43 AM
ocmyface

i feel like a caveman :stick:

i get all excited when i can hit 2.3 on my desktop 2500+ barton...
11-13-2004, 08:04 AM
OPPAINTER

Quote:

Originally Posted by FUGGER

No LN2. I picked up a new X800XT and I will put the PE bios on it.

Got lucky af Frys, they had the Visiontek X800XT in stock and it has 1.6ns ram and Rage Theatre chip on it.

Ill try not to crush the core this time ;)

My 3.8E E0 ES will be here monday but I will still get in some runs on the 3.6E E0 in the mean time.

Adding a 600w OCZ PSU to the mix

3.6E E0
DFI 875P-T
Visiontek X800XT modded
OCZ Powerstream 600w
OCZ Rev2 TCCD
OCZ DDR Booster

And just for the fun of it all I may abuse a BFG6800U as well.

Get some 3D scores up on the orb, lets see what that bad boy will do at 6gigs.

OPP
11-13-2004, 08:36 AM
eva2000

Quote:

Originally Posted by FUGGER

No LN2. I picked up a new X800XT and I will put the PE bios on it.

Got lucky af Frys, they had the Visiontek X800XT in stock and it has 1.6ns ram and Rage Theatre chip on it.

Ill try not to crush the core this time ;)

My 3.8E E0 ES will be here monday but I will still get in some runs on the 3.6E E0 in the mean time.

Adding a 600w OCZ PSU to the mix

3.6E E0
DFI 875P-T
Visiontek X800XT modded
OCZ Powerstream 600w
OCZ Rev2 TCCD
OCZ DDR Booster

And just for the fun of it all I may abuse a BFG6800U as well.

3.8E ES is :cool:
11-13-2004, 09:01 AM
IntelInside

Quote:

Originally Posted by pik-ard v1.1

IntelInside, i think the problem is, is that its only been optomized for ONE type of processor, so far as i know, theres no SSE2 optimization, MMX, 3DNOW, etc... so optomizing a program that is widely used in a community for just ONE of the contenders is unfair. what would have made people more agreeable to your modification, is if you got togeather with people who had other types of cpu's, and created a version of superpi with optimizations for ALL of them.

Using SuperPI as a benchmark which is faster on AMD because it has faster FPU is unfair.

Quote:

Originally Posted by pik-ard v1.1

the point is that the optimization is only an optimization for a certain processor type, which renders it uncomparable to other processors.

I do not agree. Why don't you run it on 386 or 486 then? Because it most likely won't run due to the program being optimized for Pentium CPU.

Quote:

Originally Posted by pik-ard v1.1

maybe you think it is useless, but thats only your personal opinion, many others think that it is a good guage of brute strength of a cpu.

NO IT IS NOT A GOOD GAUGE.

It only stresses FPU and because AMD has faster FPU it can only appeal to AMD owners as a good gauge because they end up as a winners.

But let me warn you, especially the owners of 64-bit AMD CPUs amongst you -- Microsoft wrote in the latest Driver Development Kit that legacy FPU code is not going to be used in Windows XP 64-bit. Only SSE, SSE2 will be used throughout the operating system, drivers and applications. Even their latest 64-bit compiler included with DDK (the one that will be part of Visual Studio .Net Whidbey when it comes out next year) does not produce FPU code anymore. You should find another benchmark which will measure SSE2 performance. Last time I checked even FX-55 was not ahead of Intel when it comes to SSE2.

Quote:

Originally Posted by CodeRed

But that definition I could run pifast for 1M places and absolutely smoke any superpi score

No you could not because we were talking about SuperPI and not about PiFast. PiFast uses entirely different algorhitm for calculation while even patched SuperPI doesn't change the way PI gets calculated. It is much like Formula 1 optimization -- you can tweak the engine as much as you want but it doesn't make any sense to compare times of two engines on two different tracks.

Quote:

Originally Posted by Xassius

people are mainly using superpi to compare the same group of cpu's...like Prescott -> another Prescott...so patching it would skew results...since you cannot distinguish between the two...

Is it? Last time I checked many people were comparing Intel and AMD CPUs (even in this thread) and that doesn't sound like the same group to me.
11-13-2004, 09:19 AM
macci

Quote:

Using SuperPI as a benchmark which is faster on AMD because it has faster FPU is unfair.

Useing a benchmark that runs better on AMd than intel is unfair? would it be unfair to use a benchmark that runs better on intel than AMD also? dont you think that Doom3 for instance is a very unfair benchmark too since it runs better on AMD? 3dmark2001 too! ;)
And for your Information SuperPi does NOT run all that much better on AMD.
for your viewing pleasure.

AMD vs. Intel has nothing to do with the fact that your patch is not very much liked among certain overclockers..
11-13-2004, 05:40 PM
IntelInside

Quote:

Originally Posted by macci

Useing a benchmark that runs better on AMd than intel is unfair? would it be unfair to use a benchmark that runs better on intel than AMD also? dont you think that Doom3 for instance is a very unfair benchmark too since it runs better on AMD? 3dmark2001 too! ;)

Obviously you didn't understand me. Let me put it this way. If your favorite football team has strong attack but weak defense will it be better than other teams because of that? SuperPI result for itself doesn't mean a $hit. There are other benchmarks that can test numerous aspects of a CPU and not just FPU unit.

Quote:

Originally Posted by macci

And for your Information SuperPi does NOT run all that much better on AMD.
for your viewing pleasure.

I would like to see them compared at stock clocks and not with some insane O/C using LN2.

Quote:

Originally Posted by macci

AMD vs. Intel has nothing to do with the fact that your patch is not very much liked among certain overclockers..

Those who don't like it do not have to use it, But they should not offend the patch nor the author because of that.

I really hope that the serious O/C community for which I have great respect will find some better benchmark.
11-13-2004, 06:04 PM
Xassius

Intel Inside's opinion: Super PI sucks...let's ruin it even more by making an indistinguishable patch!!!!

If you don't like it...doesn't mean other people don't like it and you don't have to ruin it for them.
11-13-2004, 06:39 PM
chilly1

a what if, what if the optimization is nothing but the use of an internal memory handling option that requires a single instruction set to operate, would the optimization be merely a correction of a deficient program>
11-14-2004, 01:08 AM
Hallowed

Quote:

Originally Posted by IntelInside

Using SuperPI as a benchmark which is faster on AMD because it has faster FPU is unfair.

NO IT IS NOT A GOOD GAUGE.

It only stresses FPU and because AMD has faster FPU it can only appeal to AMD owners as a good gauge because they end up as a winners.

So, by your definition, 98% of todays games are poor gauges of performance because AMD is faster?

If Intel were faster would it be an appropriate gauge of performance then?

Your patch functions as a crutch. It assists the needy. A true benchmark has an equal challenge / test / load on all parties. And thats all there is to it.
11-14-2004, 04:43 AM
xsky

@Fugger
Did you have any issues getting the dfi to work with those temperatures. I've good contact and the cpu is cold but the board won't boot
11-14-2004, 05:04 AM
blinky

Quote:

Originally Posted by IntelInside

Using SuperPI as a benchmark which is faster on AMD because it has faster FPU is unfair.

cry me a river dude,

life is unfair,

theres plenty of benches that love intel more than amd, no one complains, they jsut use intel for those
11-15-2004, 12:31 AM
UNKNOWN

Get Ready: P4 Up To 6.5 GHz

Hi Guys,

I’m on my way to reach 6.7GHz full info & pictures will be available soon, and who knows maybe a video :) :D

Congratulate for all the scores which you ALL have achieved

See you soon.
11-15-2004, 12:45 AM
charlie

Quote:

Originally Posted by UNKNOWN

Hi Guys,

I’m on my way to reach 6.7GHz full info & pictures will be available soon, and who knows maybe a video :) :D

Congratulate for all the scores which you ALL have achieved

See you soon.

Hahahahahaha!!!!!!!!!!
OOOhhh that's soooooo mysterious sounding! 6.7 huh?
11-15-2004, 12:46 AM
Xassius

NICE unknown...what CPU 3.8 E0 Engineering Sample? What Mobo?

regardless... 6.7Ghz would be plain sweet....(not that Fugger's 6.3Ghz isn't) :)
11-15-2004, 06:27 AM
blinky

Quote:

Originally Posted by UNKNOWN

Hi Guys,

I’m on my way to reach 6.7GHz full info & pictures will be available soon, and who knows maybe a video :) :D

Congratulate for all the scores which you ALL have achieved

See you soon.

r u in Dubai or Abu Dhabi?

Show 100 post(s) from this thread on one page