SuperPi SEE2 Patch Standardization

I know this is a hotly contested subject, and after reading some posts about it i felt that we need to come to a conclusion as to what were are going to do about the super pi patches.

Here is my thinking on the subject. At first I thought it was a bad idea as the AXP's didnt have see2, so it gave intel and a64 a big advantage (1 second). But more recently I honestly havnt been seeing much in the way of AXP's doing anything special or doing any benchmarks that can compete with anything that is on the market. And now AMD is stopping production of the AXP's. So as much as it pains me to say it, I think we need to move on from our beloved AXP's.

So I think that all of the a64 super pis should be patched for see2 from now on. It eliminates the guessing game and the alegations that somone is cheating becuase they get a better score than your unpatched version. If we all use the patch then it really wont be a "cheat" anymore. It will be inherent in the benchmark so no one will be cheating.

Let me know what you guys think of doing this. Just to let u guys know, I personally do not even have the patch nor have attempted to download it, so I am not trying to justify using the SEE2 patch for myself.

Thanks, MaxxxRacer

P.S: Guys keep it clean please, no flamming or wild allegations of any sort. I just want to hear what you guys have to say on the issue of standardizing this patch.

wot about just makin a patch too change the title of the window to "Super PI SSE2"
that way ppl can use that 4 screenshots, a simple hexedit should do it.
I realize ppl can cheat by simply not apply the title patch, but if u wanna cheat there will allways be a way to do it...
maybe XS could host a patched version with the rite title.

If ppl wanna show a unpatched run, the simply show both run, and use the difference as a validate 4 the unpatched run.

i like that idea too..

But if you can change this program to show "Super PI SSE2" a lot of people can change the code so you couldn't trust any score anymore. The good thing about the old superpi was that some guy made this program that nobody knows just for some simple things and it ends up being used a lot but nobody knows the code. We just have to find a new program thats cheatfree and a good cpu bench.

http://homepage.mac.com/v1.42ghz/pc/SPI_SSE2.jpg

somethin like this maybe?
Due too my limited hexedit skills Im stuck with the original text length... :help:

Quote:

---

-SCREENSHOT-

somethin like this maybe?
Due too my limited hexedit skills Im stuck with the original text length... :help:

---

Just imagion how easy it would be to cheat, I think the program needs some more protection against cheaters.

Would the SSE2 patch even do anything for the Athlon 64? I always thought SSE2 scales with processor speed. Someone correct me if I'm wrong. My thinking is that a 3.2ghz or beyond P4 would always be faster because SSE2 scales with clock speed.

if you can hexedit the program, then so can everyone else with a copy of Hex Workshop.

I've googled and site searched... does anyone have a LINK to sse2 patched sPi for A64>?

C

i too cant find the patch as i want to do some testing with/without the patch on my a64

It's great idea, yes. However the initiative has to come from the makes of Super PI. Possibly XS could get together to remake the Super PI to make sure we dont have 15 year old kids with photoshop skills :rolleyes:

Quote:

---

Originally Posted by charlie

I've googled and site searched... does anyone have a LINK to sse2 patched sPi for A64>?
C

---

Here you go...
I attached both patches, for SSE2 and SSE3.

Later today I'll make another edited version for the SSE2 instructions. As you can see by my first one I was able to edit the title beyond the original limit. (I screwed up the internal tables, but that will only make it hard for the next person who wants to edit it, the end user functionality is the same)

Quote:

---

Originally Posted by charlie

I've googled and site searched... does anyone have a LINK to sse2 patched sPi for A64>?

C

---

Yea it would be good to show patched and unpatched runs. Just try to not mix them up :p:

Here is the SSE2 version, just like the other one I did this one uses a seperate database from the prescott one and the normal one to keep the results seperate.

is there an SSE(1) patch...or any other patches are SuperPi for that matter

jjcom

Quote:

---

Originally Posted by jjcom

is there an SSE(1) patch...or any other patches are SuperPi for that matter

jjcom

---

no, at least not made by Ivan Levicki ;) no optimisations for AXP...

Quote:

---

Originally Posted by bachus_anonym

Here you go...
I attached both patches, for SSE2 and SSE3.

---

thx for that...i just tried it and ran 1M twice patch and unpatched...4sec difference :cool:

I only got a second difference! 32 unpatched 31 with.

i think that is about what everybody gets.

ooh, SSE2 patch got me down to 31sec :)

i only saw a 2 sec difference so i dont under stand why its such a big deal :/ could be just me tho :/

if this patch Super pi is all about recompiling the code using a better compiler, then I think those with Prescott will have the edge because most compilers are optimized for intel architecture. Intel compiler especially.
A fair SSE2 patch for A64 with code recompilation, in my opinion can only be done if the code is re-compiled with an optimized compiler for A64 (eg PathScale compiler). PathScale compiler got alot of boost when used for SPEC INT and SPEC FP benchmarks.

2 seconds in superpi1m is a heck of alot... that is like 200mhz on ur cpu...

Quote:

---

Originally Posted by Juliette

It's great idea, yes. However the initiative has to come from the makes of Super PI. Possibly XS could get together to remake the Super PI to make sure we dont have 15 year old kids with photoshop skills :rolleyes:

---

Agreed.
That's the only way I think - other than completely abandoning SuperPi that is.
Still there is no "un-hackable" code so sooner or later we'll see hacks for the new version as well.
I was thinking that the addition of a special string of digits that can be verified when we are dealing with an unpatched version will help (CPUz style).

Quote:

---

Originally Posted by craig588

Here is the SSE2 version, just like the other one I did this one uses a seperate database from the prescott one and the normal one to keep the results seperate.

---

Is THIS the one for A64?

@Charlie

The original patches made by Igor Levicki ( SSE2 and SSE3) are attached to my post somewhere above...

craig588 "hacked" the SSE2 version so it shows on the titlebar that it's SuperPi SSE2 instead of just SuperPi.