• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

adding Folding benchmarks to reviews

T

The_SLOMOWERS

n00b
Joined
Aug 28, 2008
Messages
41
Was just thinking that it would be a good idea , if the video card reviewers could add the folding benchmark figures to their reviews.
Would help to make it a lot easier in researching respective cards for folding duties.
 
The_SLOMOWERS said:
Was just thinking that it would be a good idea , if the video card reviewers could add the folding benchmark figures to their reviews.
Would help to make it a lot easier in researching respective cards for folding duties.
Click to expand...

Brother/sister, while I think you have a good idea, I just think any benchmark or said folding abilities would be bias toward the individual or company doing the benchmark (BS us :eek:)

Personaly, I don't care if the results were made by an independant person/firm/manufactuer/"poo poo" artist (PC :p) and they swear on a stack of bibles (they could be Muslim, I ain't knocking Muslims) I would only trust the word of some of the members on this forum or a few unbias individuals (like unclewebb over at XtremeSystems, etc)

Yeah, I do sound a little on the paranoid side, but then almost 60 yrs old (9 days and a wake up, but then who's countin') I am a firm believer in "it's much better to be safe (and be careful who you trust), than sorry (get bullpooped, PC :p)

Keep on a' DC'n for the cure

 
I have thought that too from time to time. I wonder how they would be able to implement it. As folding gets more and more prevalent the numbers would really start to mean something to buyers maybe. But, I just don't think they are going to want to put the time into adding folding to the review, and really, unlike CPU, folding isn't quite as good of a measure of how good the card is, just the stream processors. CPU folding taxes the whole processor for the most part, so you can't just have a lopsided over doped processor good at folding and little else.

Maybe we could get Kyle to try it on a review or two though.


 
The problem with the benchmarks is that they would have to be based on stock settings. This is [H] and we OC the crap out of everything!!! And as we all know, OCs are going to vary from person to person.

While the "stock" settings will provide us some ideas as to what the card can do, this also is misleading. As Stanford is always adding optimizations, the benchmarks themselves will become dated before they get to the general public.

 
capreppy said:
As Stanford is always adding optimizations, the benchmarks themselves will become dated before they get to the general public.
Click to expand...

This is true. I didn't even think about that. Damn. Well, it's a great idea on paper, but the implementation of it just doesn't seem like it would work well. If only someone could control and use the same work units etc. for every test, maybe that would work.

 
capreppy said:
The problem with the benchmarks is that they would have to be based on stock settings. This is [H] and we OC the crap out of everything!!! And as we all know, OCs are going to vary from person to person.

Click to expand...

True enough but there are enough different factory OC cards from pretty much all the vendors to at least be able to get some idea of the cards capability. How often do you see stock/reference cards reviewed? ;)

I don't think it's a bad idea at all but how many reviewers would spend the 20-30 minutes required on a WU to "normalize" the PPD results? I have seen a couple of F@H benchmarks done in reviews but not for GPU. The target audience just isn't large enough to support doing it I don't think.


 
Another wrench in the gears is the fact that each WU vary in PPD so it would be very complicated to have a good benchmark. The purpose of a benchmark if to have something to compare with a common ground and unless you rehash the same WU over and over, it won't work because the deadline will stop it.

 
capreppy said:
The problem with the benchmarks is that they would have to be based on stock settings. This is [H] and we OC the crap out of everything!!! And as we all know, OCs are going to vary from person to person.

While the "stock" settings will provide us some ideas as to what the card can do, this also is misleading. As Stanford is always adding optimizations, the benchmarks themselves will become dated before they get to the general public.

Click to expand...

capreppy, you're so right :eek:. This is the [H]orde and most members do OC' the "living poop" (more PC :p) outa' their gear. Stanford is always optimizing and improving the clients/WU's (although at times it seems like they'd have to speed up to stop) So in my humble opinion, the ink wouldn't even be dry on the benchmark figures before the figures were outdated (inaccurate) :eek:

@Wheresatom, I think the idea of a word from Kyle would be trusted. I think he would gather results not only from his personal data and tests but the test results from some of the very knowledgeable and informative members on the [H]orde.

Ain't we lucky most [H]orde members freely pass these "pearls" of information on to us less knowledgeable computer jocks (men and women) :p

EDIT: Sheech :eek: You see what I mean by "more knowledgeable" [H]orde members ? While I was "one finger" typing up this post brother Xilikon posts another couple of very obvious facts, the WU's have different point awards and how can you benchmark a GPU
WU that has such short time constraints :rolleyes:

Crunchin' away for the CURE

 
To elaborate on my previous point, I do believe that if you can gather all the cards and do a shootout, having a F@H benchmark would work since we can use the same basis for comparison. Using F@H in a review of a given card or cpu won't really work for the reason mentionned above.

 
One benefit might be that NVidia / ATI see Folding is being used as part of buying decisions and they may help optimise the drivers for their cards.
 
It would also have the added effect of people seeing it in the benchmarks and wonder what it actually is as well.
Which would make more people aware of it, which would be a good thing.
And yes the benchmarks would need to be done on a non over-clocked card.
 
The only way I could see something like this working is if you ran 2 benchmarks.
One with a standard slow protien and one with a standard fast one.
That would give you the approx min and max PpD from the card.

Then overclock the hell out of the card and rerun the benchmarks ........ :p

Also the ATI client is much more CPU bound than the nVidia client.
This would make chooseing the right benchmarking hardware more tricky.

Luck ......... :D
 
I would like to think that someone, here in this brain cell laden [H]orde could figure out how to save a WU for repeated execution on GPU's, even if not a critical WU, just for sake of repeatablity. Maybe a simple and complex WU for sake of diversity??

Still, points noted about optimizations and relevancy, it seems like a losing challenge. For every obstacle we overcome, more relevant changes occur. Sounds like folding challenge to me...

At least more pronounce mention of shaders/bus width potential folding performance (even if only calculated).
 
The easy way to make a benchmark protien is just to make a copy of your folding folders at the start of a Protien and your UserID.

Then to benchmark .........
Add the UserID to the registry if needed. This way no internet access is required
Copy the folding folders to the computer.
Make a shortcut pointing at the client and the data.
Start and let it run 10-20 frames.
To rebenchmark just recopy the data back into the box.
Easy .............:p

Edit. You also need to config the client to ignore deadlines.

Luck ............... :D
 
It sure would be nice if someone on the beta testing team got one of the Stanford developers to come up with a testing protein for us to benchmark with in their spare time. ;) /sarcasm

I know there is not time at all for this, but it would be cool. Maybe they would make a standalone client that was only good for testing GPU and CPU combos.

Gotta get the needs finished before the wants though.

 
From HighYield in another posting "Even if everyone overclocks which certainly doesn't happen, how would the Stanford group be able to reasonably predict how their WU will work on X number of different MOBOs, Y different GPUs, Z different power supplies, and N different cooling solutions all with variable levels of OC and extra voltage applied? Wouldn't the best we can hope for is the WUs work ok on stock boxen? Maybe an actual programmer could comment on this (I am not...unless A/C Basic on an Amiga 20 something years ago counts )"

Yeh, sure.
 
This subject has come up numerous times in the past and each time Kyle has rejected the idea. I think the biggest issues have been covered by everyone above. Simply put the results would be conflicting on various levels and Kyle would spend his days wasting his time to explain the whys and wherefores.

I can put a video card into a 680i based mobo and get different results using the exact same card on an I-35 board and an I-45 board. Then you run into Vista-64 being a bit faster than XP anything.

I just don’t see it as a practical solution to accurately anyone’s question on “fastest”.;)


 
jws2346 said:
Brother/sister, while I think you have a good idea, I just think any benchmark or said folding abilities would be bias toward the individual or company doing the benchmark (BS us :eek:)

Personaly, I don't care if the results were made by an independant person/firm/manufactuer/"poo poo" artist (PC :p) and they swear on a stack of bibles (they could be Muslim, I ain't knocking Muslims) I would only trust the word of some of the members on this forum or a few unbias individuals (like unclewebb over at XtremeSystems, etc)

Yeah, I do sound a little on the paranoid side, but then almost 60 yrs old (9 days and a wake up, but then who's countin') I am a firm believer in "it's much better to be safe (and be careful who you trust), than sorry (get bullpooped, PC :p)

Keep on a' DC'n for the cure

Click to expand...

Almost 60 huh? Gotcha beat by a year although as life goes on this becomes one contest I wish would slow down;)



 
BillR said:
Almost 60 huh? Gotcha beat by a year although as life goes on this becomes one contest I wish would slow down;)



Click to expand...

Hey BillR, you figure out how to "rig" that contest and you'll be more famous than Howard Hughs was (and be a whole lot richer too) :eek:

DC for the CURE

 
Pooch666 said:
From HighYield in another posting "Even if everyone overclocks which certainly doesn't happen, how would the Stanford group be able to reasonably predict how their WU will work on X number of different MOBOs, Y different GPUs, Z different power supplies, and N different cooling solutions all with variable levels of OC and extra voltage applied? Wouldn't the best we can hope for is the WUs work ok on stock boxen? Maybe an actual programmer could comment on this (I am not...unless A/C Basic on an Amiga 20 something years ago counts )"

Yeh, sure.
Click to expand...

Sure about what? Care to elaborate? That was posted for a differernt thread.
http://www.hardforum.com/showthread.php?t=1344039&page=5
 
Maybe benchmark was the wrong word ie
Something along this card did x whilst folding for this review.
Once people see something like that they would naturally wonder what folding is.
As they would looking into what folding is and maybe start folding themselves.
Maybe look at it as a way to get more people involved in folding for a cure.
 
You must log in or register to reply here.
Back
Top