Go Back   nV News Forums > Hardware Forums > Benchmarking And Overclocking

Newegg Daily Deals

Reply
 
Thread Tools
Old 11-11-03, 03:35 PM   #85
vandersl
Registered User
 
Join Date: Mar 2003
Location: NJ
Posts: 43
Default

Quote:
I dont see the difference with having a compiler in a driver making shaders work better, or a driver engineer making shaders work better. Ideally the former is preferred, but until they get a miracle compiler built, it aint gonna happen.
Quote:
The whole issue of benchmarks is just totally flawed. I thought everyone agree'd ages ago to stop paying attention to PenisMarks.. sorry 3dmarks?
There's a HUGE difference.

In one case, there is some probability that when a new game comes out, the increased performance will be seen in that game. After all, that's what a proper benchmark is all about - predicting performance in other applications.

In the other case, the performance gain was 'faked', and the new game won't benefit.

Now, unless you trust your IHV enough to believe they will continue to dedicate engineering talent to optimizing future games for your hardware, you should care about the difference. Saying you don't care is somewhat short-sighted.

There seems to be two camps on this issue. One can see that cheating on benchmarks does no one any good except the IHV doing the cheating. The other camp just says 'I am smart enough to realize that benchmark X means nothing, and will base my buying decisions on games, so I don't care if they cheat <yawn>'.

I don't think there's much hope for converting people in one camp to the other, and I certainly won't try. What bothers me most is when the people in the second camp use a condescending tone in their arguments, implying that to care about benchmark integrity is somehow 'nerdy', or a waste of time. It's particularly ironic when they say 'don't waste your time - play games instead'.
vandersl is offline   Reply With Quote
Old 11-11-03, 03:39 PM   #86
Nutty
Sittin in the Sun
 
Nutty's Avatar
 
Join Date: Jul 2002
Location: United Kingdom
Posts: 1,835
Send a message via MSN to Nutty
Default

Quote:
I can't believe you just said that...that has to be the stupidest thing I have ever read.
I try.

Quote:
A benchmark is supposed to be fair and impartial...an apples to apples comparison or as close to one as is possible. What you are suggesting is an apples to oranges comparison where oranges are the preferred fruit. I hope I made this simple enough for you to see your error.
I see what you're getting at, but its not valid anymore. _ALL_ future games will have their shaders tinkered with by a much more complex compiler than we currently have. This will be true for all IHV's. Not just the ones that balls up their hardware. Putting a patch out that deliberately circumvents these optimizations is wrong.

Quote:
And about those compilers...I have no problem with them at all so long as they don't decrease image quality or do stuff that wasn't intended by the developer. If a developer asks for output A using X instructions the compiler should optimize X instructions but still output A. Anytime that output A is different because of a compiler should raise red flags.
Okay, so what if a driver engineer manually changes the shader so that it still outputs A, but is loads faster for their hardware? Is that wrong? To me thats good. I'd prefer a bit of automation in there, but NV wrongly decided on how much investment to put into shader compilers for their current hardware. The end result is they had to manually do it, or die.

Quote:
Tests like the one here with 3DMark show that it sure as hell wasn't the compiler providing the optimization but hand picked "optimizations" instead. A compiler would not suddenly lose performance in a program because that program changes itself so it could no longer be detected by cheating drivers.
True, but you said if it outputs the right stuff, with the same IQ by a compiler its fine. NV are just manually doing what their future compiler should do.

Quote:
NVIDIA has definately lost me as a customer till at least the NV50...and that is provided they remove all their old cheats and can show they have a IQ and perfomance lead on ATI.
Disregarding AA quality, what area does NV fall behind in? I personally think NV has better image quality. Their texture sharpness always looks that little bit better to me.
Nutty is offline   Reply With Quote
Old 11-11-03, 03:43 PM   #87
titan
Registered User
 
Join Date: Nov 2003
Posts: 10
Default

hey FX guys

Here are my results tested with my FX5950, 256mb,DDR


Patched with the 3.30 version




Patched with the 3.40 version




greetz

TiTaN
titan is offline   Reply With Quote
Old 11-11-03, 03:47 PM   #88
Nutty
Sittin in the Sun
 
Nutty's Avatar
 
Join Date: Jul 2002
Location: United Kingdom
Posts: 1,835
Send a message via MSN to Nutty
Default

Quote:
There seems to be two camps on this issue. One can see that cheating on benchmarks does no one any good except the IHV doing the cheating.
All the more reason to ignore benchmarks that can easily be manipulated.

Quote:
implying that to care about benchmark integrity is somehow 'nerdy', or a waste of time.
It is a waste of time IMO, if its not done correctly.

A benchmark should be; "How fast does it do this task".
3dMark on the other hand is; "How fast does it do this bit of code, without the option to massage it for the hardware".

They are not the same. If the code isn't particularly suited to the hardware, its gonna look bad. But there is always more than 1 way to do a particular task. And it should be upto the driver/compiler how it implements that task, to make use of its hardware the best.

Thats my take on it, if I appear condescending sorry, but I'm all for driver massaged shaders. IHV's know the quirks and performance of the hardware better than developers, so leave it upto them to get the code running right. As long as IQ is not compromised its all good.
Nutty is offline   Reply With Quote
Old 11-11-03, 03:50 PM   #89
ChrisW
"I was wrong", said Chris
 
Join Date: Jul 2002
Location: standing in the corner!
Posts: 620
Default

Well, at least the scores did not plummet like they did with the last anti-cheat patch. Only an 800 point drop means they have made lots of legitimate optimizations this time (or some of them are now considered legitimate).
__________________
AIW 9700 Pro | 1.3GHz CeleronT | 512MB PC133 SDRAM | ECS PCIPAT Mobo (Intel 815EP)
RadLinker/RadClocker
ChrisW is offline   Reply With Quote
Old 11-11-03, 03:55 PM   #90
cthellis
Hoopy frood
 
Join Date: Oct 2003
Posts: 549
Default

Quote:
Originally posted by Nutty
The point is, they wouldn't. Unless IQ is being sacrificed. If its not, then wheres the problem?

I mean I dont care that NV is cheating if the IQ stays the same. I really dont.
Because unless you examine a game closely and make comparison, you won't know when, where, and how IQ gets sacrificed, and you won't know what future areas it might break. What might get broken if the developer makes a core patch or introduces new maps, or modders make their own variants on an engine while unaware of different ways IHV's approach it? And considering how many ways IHV's could optimize only for game benchmarks--without compromizing IQ--how can you trust the numbers OR the quality in comparison to the way the actual game will perform? FRAPS is not repeatable enough yet, and not every game allows one to record their own runthroughs to use FRAPS on, so for right now these kinds of worries are entirely valid. How do you get proper analysis of the future (something very important to consider for any smart consumer) off things we can only see now and can't trust their whole process--the important part--and can just see the visual end-results?

Basically, though the ABSOLUTE LAST step may not really matter in this case, all the intermediary ones can have undesirable side-effects, and would just seem to be the wrong paths to reward. It's fragmenting even more an area of the computer world that's hard enough to follow, and that we've spent many years trying to SOLIDIFY.

---

On Futuremark's current steps, I think it's definitely the best approach to take, since simply labelling a current set of drivers as "cheating" and forcing results from a few drivers back can kill many valid optimizations as well as the invalid ones. If Futuremark is confident they excised everything objectionable to their new rules, then they are free to approve the current drivers under their new patched version. It removes offending paths while keeping valid ones, as well as being the "smoothest" adoption method. The critical steps are the ones that go from here on out. They've been VERY careful in working up their guidelines and parameters so as to leave no wiggle-room, and they've "reset the counter" as it were. How will they respond if nVidia were to patch 52.16 to defeat their current 3DMark2003 build? How will they respond to later infractions from nVidia, ATi, or any other source?

They've taken a solid stand on THEIR desires alone, and made a long, careful progression to this point, but the maintainance will likely be even harder and more critical to ensure they don't have to do this again.
cthellis is offline   Reply With Quote
Old 11-11-03, 03:55 PM   #91
Blacklash
8^9^3
 
Blacklash's Avatar
 
Join Date: Oct 2003
Location: Old Vizima
Posts: 3,679
Default r

well thats my old score,

http://service.futuremark.com/compare?2k3=1514656

before I took the 730pnt hit

someone wanted details<
__________________
Intel Core i7 920 @ 3.96GHz (1.36v)|Mushkin 998681 XP3-12800 (3x2GB)
ASUS TUF Sabertooth (X58)|ASUS GTX 580 DirectCU II (980|4604)|ASUS PA246Q
WD VelociRaptor 150GB HD (x2)|Pioneer DVR-2920Q|LG GH22LS30|Klipsch PM20 2.0
SilverStone OP1000-E|SilverStone TJ10-B|Thermalright U-120 Extreme|Win 7 HP x64
Blacklash is offline   Reply With Quote
Old 11-11-03, 03:56 PM   #92
nForceMan
 
Join Date: Apr 2003
Posts: 330
Default

Quote:
Originally posted by titan
hey FX guys

Here are my results tested with my FX5950, 256mb,DDR

...

greetz

TiTaN
Have you tried it with ForceWare 52.20 as well?
nForceMan is offline   Reply With Quote

Old 11-11-03, 03:57 PM   #93
ChrisRay
Registered User
 
ChrisRay's Avatar
 
Join Date: Mar 2003
Location: Tulsa
Posts: 5,101
Default

Quote:
Originally posted by nForceMan
Have you tried it with ForceWare 52.20 as well?
Eh? Where are these drivers?
__________________
|CPU: Intel I7 Lynnfield @ 3.0 Ghz|Mobo:Asus P7P55 WS Supercomputer |Memory:8 Gigs DDR3 1333|Video:Geforce GTX 295 Quad SLI|Monitor:Samsung Syncmaster 1680x1080 3D Vision\/Olevia 27 Inch Widescreen HDTV 1920x1080

|CPU: AMD Phenom 9600 Black Edition @ 2.5 Ghz|Mobo:Asus M3n HT Deluxe Nforce 780A|Memory: 4 gigs DDR2 800| Video: Geforce GTX 280x2 SLI

Nzone
SLI Forum Administrator

NVIDIA User Group Members receive free software and/or hardware from NVIDIA from time to time to facilitate the evaluation of NVIDIA products. However, the opinions expressed are solely those of the members
ChrisRay is offline   Reply With Quote
Old 11-11-03, 03:58 PM   #94
bkswaney
Mr. Extreme!
 
bkswaney's Avatar
 
Join Date: Aug 2002
Location: SC
Posts: 3,421
Send a message via Yahoo to bkswaney
Default

Quote:
Originally posted by ChrisW
Well, at least the scores did not plummet like they did with the last anti-cheat patch. Only an 800 point drop means they have made lots of legitimate optimizations this time (or some of them are now considered legitimate).
I agree.

I just sent a nice email to everyone at nvidia about this.


Has anyone tried the olders CAT drivers with this new patch.
ATI just released the cat 3.9 a few days ago.
I wonder if they knew this patch was coming.
__________________
Notebook!
Compaq Presario CQ60-215DX
AMD 64 Athlon X2 @ 2GHz (QL62)
15.6 inch HD WideScreen
Nvidia 8200M-G 895mb
2Gig system ram
250Gig SATA 5400rpm HDrive
Vista Premium
bkswaney is offline   Reply With Quote
Old 11-11-03, 04:01 PM   #95
Ady
...
 
Ady's Avatar
 
Join Date: Nov 2002
Location: Australia
Posts: 502
Default

i've only seen cat 3.8 & 3.9 tested.
__________________
Dying is not going to kill me.
Ady is offline   Reply With Quote
Old 11-11-03, 04:01 PM   #96
fallguy
 
fallguy's Avatar
 
Join Date: Oct 2003
Location: ^ Next to her
Posts: 155
Default

Quote:
Originally posted by ChrisW
Well, at least the scores did not plummet like they did with the last anti-cheat patch. Only an 800 point drop means they have made lots of legitimate optimizations this time (or some of them are now considered legitimate).


800 down from 5300 is a pretty large percentage....
fallguy is offline   Reply With Quote
Reply


Thread Tools

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Forum Jump


Similar Threads
Thread Thread Starter Forum Replies Last Post
NVIDIA Driver 295.53 installs into kernel 3.4 without a patch! jdmcdaniel3 NVIDIA Linux 3 06-08-12 09:41 AM
Need Help Installing NVIDIA Tesla M2070Q in Linux RHEL5 Ferianto85 NVIDIA Linux 0 05-18-12 08:35 PM
Rumor regarding lack of 680 availability ViN86 Rumor Mill 6 05-09-12 04:48 PM

All times are GMT -5. The time now is 11:36 AM.


Powered by vBulletin® Version 3.7.1
Copyright ©2000 - 2014, Jelsoft Enterprises Ltd.
Copyright 1998 - 2014, nV News.