Sign in with
Sign up | Sign in
Your question

Crysis Warhead!!Ati Rock with 8.9 Driver!

Last response: in Graphics & Displays
Share
September 19, 2008 3:57:54 AM

The 9800 GTX+, amd 4850, amd 3870, amd 4870, 8800 GT, 8800 GTS, GTX260, have higher minimum framerates than the GTX 280

These benchmarks are total hogwash, I can appreciate your enthusiasm but its pretty apparent they aren't repeating the test runs per card to find an average - which is pretty poor practice.

I'm not trying to be rude to you, but i'm just pointing out that this review is obviously quite biased and unrealistic.
September 19, 2008 4:03:37 AM

I have seen other benchmarks that showed the 280 GTX dipping to a much lower minimum fps than those cards, so something is probably going on, but not extreme enough to justify those numbers.
Related resources
September 19, 2008 4:10:17 AM

ovaltineplease said:
The 9800 GTX+, amd 4850, amd 3870, amd 4870, 8800 GT, 8800 GTS, GTX260, have higher minimum framerates than the GTX 280

These benchmarks are total hogwash, I can appreciate your enthusiasm but its pretty apparent they aren't repeating the test runs per card to find an average - which is pretty poor practice.

I'm not trying to be rude to you, but i'm just pointing out that this review is obviously quite biased and unrealistic.


I'm not saying that the benchmarks are not bad, but ovaltine you are some what biased to the Ovaltine/nvdian fraction conglomerate [read into it, it's a fact! :D  ] ...so your comments carry that undertone, no matter how much you try to clear that up. :pt1cable: 

It's so bad now that I automatically think Ovaltine is part of "The It's Meant To Be Drunk" camp...DAMN you ovaltineplease, damn you!!!! :lol:  :lol:  :lol: 
September 19, 2008 4:16:28 AM

I own an AMD 4850 in my second PC, I don't really know how using Nvidia gtx260s in my main gaming PC makes me biased.

Anyways, I know you don't mean anything by it - but that website is proven amateur; they can post whatever they want but I would respond no differently if they posted a 4870x2 result having a lower min fps than a gtx280 in any other game - it would likewise be totally unbelievable nonsense.

ed: one thing I should clarify - i'm only an EVGA fanboy, and only cause of their services : O~
September 19, 2008 4:31:55 AM

Ovaltine, on an unrelated note, what did/are you doing with your 8800GTs
September 19, 2008 4:56:27 AM

in no way are these biased [unless they are fake].. if they did one round with each card, and got these results from the one round, than hey, it gotta be real. it could even be luck. can't be biased though

*edit*
on another note, this site doesn't seem to have any credibility, so my last comment is worthless. lol

*another edit*
my proof: no author. of course, one of the first things you learn in hs/college is to never read/believe anything that doesn't have the presence of an author's name.. if psycho's the author, well that just makes my point a little more substantial
a b U Graphics card
September 19, 2008 5:17:26 AM

eklipz330 said:
in no way are these biased [unless they are fake].. if they did one round with each card, and got these results from the one round, than hey, it gotta be real. it could even be luck. can't be biased though

Biased, no. Inconclusive, yes. A single benchmark run can't be trusted. I've done them myself with Crysis, and you will often get different framerates the second time round.
a b U Graphics card
September 19, 2008 10:30:04 AM

4850 ties with a gtx260 looks fake?
ati cards dont run well on crysis compared to nvidia cards, due to optimization.
September 19, 2008 10:48:53 AM

Yea, exactly; anyone who has done any benching on Crysis (or really any games modern games) will have to do 2-3 runs in order to get a conclusive result as randomizer said.

I'm not against gameplay benchmarking, hardocp frequently uses this method - however you have to do a LONGER run than 40 seconds, that is just not long enough to make a conclusive benchmark.
a b U Graphics card
September 19, 2008 10:50:48 AM

Ive seen the G280 on various sites using various benches showing lower minimum fps. Its not bias, its reality. Look for yourselves
a b U Graphics card
September 19, 2008 10:54:30 AM

The reason I remember this is because I was surprised by it. Its an exception, not the rule, but it happens, and more often than you think
September 19, 2008 12:38:57 PM

If I run my gtx260 in non-sli I don't get those min framerates on a fraps benchmark - its not reality; I could accept a 1 fps difference, but not between 5-10 - thats just laughable.

I mean good god the 9800GTX+ framerates:

min 15, avg 17.


lol?
September 19, 2008 1:05:24 PM

ovaltineplease said:
If I run my gtx260 in non-sli I don't get those min framerates on a fraps benchmark - its not reality; I could accept a 1 fps difference, but not between 5-10 - thats just laughable.

I mean good god the 9800GTX+ framerates:

min 15, avg 17.


lol?


um 5-10 is completely possible given different hardware/software set-ups, you'd be surprised how the smallest things can affect fps...
September 19, 2008 1:10:12 PM

W/e. I don't intend to argue the point as I already know this website is a bunch of amateurs to begin with, I would suggest waiting for a better website to do a technical review on the game cause this "review" is a jokeshow.
a b U Graphics card
September 19, 2008 1:24:07 PM

Ive seen more than 1fps, but 10, no. Im thinking they just gave lowest read, even for a single frame. Thats doesnt mean itd be noticeable, tho . Could be on fireup. All Im saying is, they arent the only ones to show a G280 with lower minimum fps than the 260 or the 4870. As we see more benches well know more
a b U Graphics card
September 19, 2008 1:27:28 PM

Theres a few people who value this site. Why do you say that?
September 19, 2008 1:29:25 PM

It's also worth noting that the 4870 isn't producing playable FPS in the enthusiast test either, so its victory there is irrelevant as far as I'm concerned.

Why is it so hard to believe that the 4870 can have higher minimums than the GTX280? The minimum probably comes when textures are streaming in, and the Radeons are know to have superior memory management.
September 19, 2008 1:31:44 PM

You might want to consider this benchmark pulled from Tom's.

http://www.tomshardware.com/charts/gaming-graphics-char...

It's not exactly the same as PCgames but very close as a benchmark. But it shows that the PCgames tests are repeatable in other sites and yield simular results. Notice that the 260 gtx is sli with 896 of mem and only does 0.3fps better? The hd4870 has 512 mb mem. The gtx 280 with 1024mb mem only does 1.1fps better. In this test only two cards beat the HD 4870 and those are the GTX280 1024mem and the 9800gx2 512x2 mem in that order. You will notice how the other Nvidia cards are sli'ed and the ATI is a solo card. As you said Crysis is optomized for Nvidia. All these top cards seem to be nearly at a tie give or take a frame. The highest performing cards will do better at higher resolutions. The sorry thing in all of this is that the very best hardware can pull only 26-29fps at 1680x1050, 0xAA, Trilinear, Very High Quality. The game coding in Crysis is poor. If you like the game great. One shouldn't have to have a 9800 gx2 tri-sli system to pull 60-70 fps on a game at 16x10 rez. I don't think that we should go to Crysis as the benchmark of record.
a b U Graphics card
September 19, 2008 1:44:29 PM

Like Ive said, you dont have to search far to find this. Slamming this site just doesnt do it for me. Looking at real numbers, other sites numbers backs up their findings. Also, for "amateurs theyre getting some nice interviews http://www.pcgameshardware.com/aid,602522/News/Exclusiv...
September 19, 2008 2:26:13 PM

topper743 said:
You might want to consider this benchmark pulled from Tom's.

http://www.tomshardware.com/charts/gaming-graphics-char...

It's not exactly the same as PCgames but very close as a benchmark. But it shows that the PCgames tests are repeatable in other sites and yield simular results. Notice that the 260 gtx is sli with 896 of mem and only does 0.3fps better? The hd4870 has 512 mb mem. The gtx 280 with 1024mb mem only does 1.1fps better. In this test only two cards beat the HD 4870 and those are the GTX280 1024mem and the 9800gx2 512x2 mem in that order. You will notice how the other Nvidia cards are sli'ed and the ATI is a solo card. As you said Crysis is optomized for Nvidia. All these top cards seem to be nearly at a tie give or take a frame. The highest performing cards will do better at higher resolutions. The sorry thing in all of this is that the very best hardware can pull only 26-29fps at 1680x1050, 0xAA, Trilinear, Very High Quality. The game coding in Crysis is poor. If you like the game great. One shouldn't have to have a 9800 gx2 tri-sli system to pull 60-70 fps on a game at 16x10 rez. I don't think that we should go to Crysis as the benchmark of record.


THG got slammed when they posted that article, and you're using it as a source reference?
a b U Graphics card
September 19, 2008 2:38:54 PM

topper743 said:
You might want to consider this benchmark pulled from Tom's.

http://www.tomshardware.com/charts/gaming-graphics-char...

It's not exactly the same as PCgames but very close as a benchmark. But it shows that the PCgames tests are repeatable in other sites and yield simular results. Notice that the 260 gtx is sli with 896 of mem and only does 0.3fps better? The hd4870 has 512 mb mem. The gtx 280 with 1024mb mem only does 1.1fps better. In this test only two cards beat the HD 4870 and those are the GTX280 1024mem and the 9800gx2 512x2 mem in that order. You will notice how the other Nvidia cards are sli'ed and the ATI is a solo card. As you said Crysis is optomized for Nvidia. All these top cards seem to be nearly at a tie give or take a frame. The highest performing cards will do better at higher resolutions. The sorry thing in all of this is that the very best hardware can pull only 26-29fps at 1680x1050, 0xAA, Trilinear, Very High Quality. The game coding in Crysis is poor. If you like the game great. One shouldn't have to have a 9800 gx2 tri-sli system to pull 60-70 fps on a game at 16x10 rez. I don't think that we should go to Crysis as the benchmark of record.


those charts seem bs to me. the gtx260 in sli get 25.20 fps on 1680x1050 no aa v. high quality.
i have my 2 gtx260 in sli and i avg the about the same fps with the same settings except i have 16aa on.
when i oc my cards i get 30-35-fps.
September 19, 2008 3:38:06 PM

I'm going to duplicate their benchmark on my system with a gtx260 in non-sli mode, overclocked

and stock-clocked with a Core2Duo processor and 4gb RAM. I am using their method, at close to

the same point and following the same methodology - starting the benchmark as soon as the game

loads while still loading textures and stopping at 40 seconds.

http://img401.imageshack.us/my.php?image=warheadsavevy2...



Tests:

GPU: 684/1404/1215, CPU: core2duo 4.05ghz, RAM: 4gb ddr2-800 unlinked 5/5/5/18

1680*1050, Enthusiast settings, NO AA/AF

Min 21, Max 28, Avg 24

http://img235.imageshack.us/my.php?image=desktopnw5.jpg



1680*1050, Enthusiast, 16XAF

Min 14, Max 22, Avg 18





GPU: 576/1242/999, CPU: 3.0ghz, RAM: 800mhz


1680*1050, Enthusiast, 16XAF

Min 14, Max 21, Avg 17



GPU: 684/1404/1215, CPU: 3.0ghz, RAM: 800mhz


1680*1050, Enthusiast, 16XAF

Min 16, Max 24, Avg 19


Suffice to say I don't expect my benchmarks to be the starting and ending authority on Crysis benchmarking nor do I find the performance mind-bogglingly impressive with a single gtx260 at "Enthusiast settings" - its clearly not playable with AF enabled at stock clocks any way you cut it.

However, the fact that my single gtx260 is vastly outperforming their gtx280 in min fps - with the only major differentiating factor being that I am using 4gb of RAM vs their 2gb of RAM prooves either 1 of 3 things: one, you need more than 2gb of RAM (obvious and proven) two, their testbed is a gongshow or three, they fail at testing without a timedemo (quite possible too)

I reserved the csv files for the other timedemos for the sake of validation if anyone cares to question it - but I thought i'd save myself a ton of trouble uploading that many images.
a b U Graphics card
September 19, 2008 6:49:07 PM

ovaltineplease said:

I'm not trying to be rude to you, but i'm just pointing out that this review is obviously quite biased and unrealistic.


Like Randomizer said, not bias so much as unreliable IMO.

As for the minimum fps issue remember an average is including very high end numbers, and the min would represent a bottleneck of whatever is there. It could be a momentary dip, and I have seen it many times with many different otherwise powerful cards.

Seeing a card with a high 70fps average but low min 10 fps (better to have median low IMO) would show variability including very high highs, but another card showing a 40fps min and a 60fps, would mean the highs aren't high, but there's also few drops where some area is bottlenecked for whatever reason. It's tough to tell from a benchmark, but they don't look biased, or fake or anything just based on that, I have more of a problem with what we've discussed many times, not running multiple runs or resolutions.

invisik said:
4850 ties with a gtx260 looks fake?
ati cards dont run well on crysis compared to nvidia cards, due to optimization.


That's the DX9 path, the Crysis DX10 path favours ATi actually.

Take not of their comments, which reflect most tests of the current Crysis: "At high details (Gamer mode) the Geforce GTX 280 is up ahead with the GTX 260 and the HD 4870 behind it... At maximal details (Enthusiast mode) the Radeon HD 4870 is the winner.."

Anywhoo, I expect Warhead to run similarly to the original Crysis, with lower shader quality running better on the GTX and the Very high shaders to run better on the HD4K series. The differences are minor and many people favour resolution over effects, so they run DX9 with the quality mods.
September 19, 2008 6:49:48 PM

^Thanks for that ovaltine. Do note that performance in this game is GREATLY improved by using DX9 mode. As far as I can tell the only thing you'll be missing out on is object motion blur, as it is not present under DX9 even with all 'enthusiast' settings.
September 20, 2008 4:24:49 AM

You really do take this personally, don't you? :( 

It's obviously some crack-head results that site got, it's obviously off, but Ovaltine you can't go after every ignorant person, you know what they say stupid is as stupid does. :pfff: 

I love how you actually went ahead and did all this just to prove how wrong that site is, that's actually pretty cool. It will help those who are researching what card to buy and are on the ropes, and for that Jen-Hsun Huang Schlong Nolong thnx you for your continued efforts to help Nvidia in these financially trying times. :D 

Anyways, don;t get all tripped out, b/c ignorance is bliss and some people can't live w/o it. :whistle: 

But keep up the fact checking, we always need more of that.
September 20, 2008 6:17:15 AM

My main concern with the review is that they use 2GB of RAM with Vista64. Although they use a system that can be considered an average gaming rig, I would assume that they were running into some bottlenecks with their RAM. If they had 4GB I bet their FPS dips wouldn't be as low.

An aspect of Warhead was to make it more optimized so that it could run better on such systems as the one tested on. I think the review yields some decent info for people with the average gaming rig looking to see how well they can run it.
September 20, 2008 11:28:42 AM

SpinachEater said:
My main concern with the review is that they use 2GB of RAM with Vista64. Although they use a system that can be considered an average gaming rig, I would assume that they were running into some bottlenecks with their RAM. If they had 4GB I bet their FPS dips wouldn't be as low.

An aspect of Warhead was to make it more optimized so that it could run better on such systems as the one tested on. I think the review yields some decent info for people with the average gaming rig looking to see how well they can run it.



Actually they have a disclaimer over the last chart. I don't know why they didn't simply correct the chart.
Quote:

Contrary to what the legend says, all benchmarks were made on Windows XP x86.
September 20, 2008 2:34:35 PM

Sheeeeeesh....nice site....I retract my statement about it being helpful.

"Contrary to what the graph says, it is actually data from an old Crysis benchmark and they were recorded at a totally different resolution but we aren't quite sure what it was..."
September 20, 2008 5:49:58 PM

I didn't take the link from an article. You can find a link to all of Tom's tested benchmark results on their main page. Point again is that at least two sites have simular results. Oh, Xbit labs very close to same numbers that makes three.

http://xbitlabs.com/articles/video/display/force3d-hd-4...

This thread is indicitive of a sytemic problem here on the forums. Some folks are just rude they just can't help themselves. Arrested devlopment? Dandy (OP) makes a post. Which was really just a link with a comment to read and enjoy and did it pleasantly. The first guy out gets nasty and then states I don't want to be rude, right after being rude, then continues to be rude . What did the OP do to cause such a reaction? Perhaps the site sighted was somewhat imperfect, but the web is full of data with very close to the same results. Some of tests vary, different cpu, mb, mem or benchies but all are close in what it is they tell us. But no someone must run their own (if they did) to tell us what is really the "truth". Who should we trust to bring us the most accurate info THG, pcgameshardware, xbit labs or well you fill in the blank__________?
September 21, 2008 12:11:40 AM

I think its worth noting that the 48xx series has been even in crysis with their nvidia counterparts on a large number of sites since day one. The only performance gap remaining in that game is with crossfire vs. sli. So if you see a 4850/70 keeping up with nvidia in crysis that's quite accurate.
September 21, 2008 12:28:40 AM

JAYDEEJOHN said:
Theres a few people who value this site. Why do you say that?

How do you put that little foto in a left corner ?
September 21, 2008 4:10:17 PM

LAN_deRf_HA said:
I think its worth noting that the 48xx series has been even in crysis with their nvidia counterparts on a large number of sites since day one. The only performance gap remaining in that game is with crossfire vs. sli. So if you see a 4850/70 keeping up with nvidia in crysis that's quite accurate.

QFT. RV770 does very well in Crysis, but Crossfire scaling at low resolutions and IQ settings is weak.
September 22, 2008 3:29:21 AM

FrozenGpu said:
You really do take this personally, don't you? :( 


No, I don't, but this is a forum which is used to help people and these benchmarks were blatantly wrong - in other words, it encourages people to buy things they don't need. For instance, if the gtx280 numbers are totally off which they obviously are - then the average user will believe that they need gtx280 SLI or 4870x2 or 4870 Crossfire in order to have adequate performance - when this is actually anything but the truth and they will likely be perfectly happy with a more midrange graphics solution unless they are playing at 24"+ resolution.

This isn't personal, this isn't Nvidia vs ATI, this is about trying to do fellow consumers a favor and making sure people are informed that some sites are simply not altogether reputable; at least in my opinion.


I love how you actually went ahead and did all this just to prove how wrong that site is, that's actually pretty cool. It will help those who are researching what card to buy and are on the ropes, and for that Jen-Hsun Huang Schlong Nolong thnx you for your continued efforts to help Nvidia in these financially trying times. :D said:

I love how you actually went ahead and did all this just to prove how wrong that site is, that's actually pretty cool. It will help those who are researching what card to buy and are on the ropes, and for that Jen-Hsun Huang Schlong Nolong thnx you for your continued efforts to help Nvidia in these financially trying times. :D 


I don't really know what you're talking about, but feel free to refer to the bunny with a pancake on its head picture.


But keep up the fact checking, we always need more of that. said:

But keep up the fact checking, we always need more of that.


I do my best, i'm admittedly not always right - but I try to be thorough when I do make a statement.
September 22, 2008 4:14:49 AM

OH MY GOD!

Look at the Nvidia Driver they are using! 177.41

Thats like 3 MONTHS OLD!

If they used the more optimized and better coded for Crysis 177.98 drivers, which actually support the GTX series unlike the .41 drivers, the GTX 260 and 280 would OBVIOUSLY absolutely DESTROY the 48XX series.
a b U Graphics card
September 22, 2008 4:49:23 AM

177.98 is really buggy. You need to drop the digital vibrance from 100% upon every restart. I reverted to 177.92 after a few hours.
September 22, 2008 12:09:17 PM

LunaticWolf said:
OH MY GOD!

Look at the Nvidia Driver they are using! 177.41

Thats like 3 MONTHS OLD!

If they used the more optimized and better coded for Crysis 177.98 drivers, which actually support the GTX series unlike the .41 drivers, the GTX 260 and 280 would OBVIOUSLY absolutely DESTROY the 48XX series.

Proof?
September 22, 2008 3:18:21 PM

It is just a pun on all the people complaining about Toms using CAT 8.6...
September 22, 2008 3:30:30 PM

Oh, cool. Thing is, at least from what I've read, Catalyst updates typically bring more to the table in terms of performance than the few and far between Forceware updates. I recently updated to the 177.92 BETA and I didn't notice a shred of difference on my 8800GT.
September 22, 2008 3:39:54 PM

Aye, but they update far less than Nvidia.

.41 was released in June, as was (pretty sure) CAT 8.6. Kind of strange they are comparing a 3 month old nvidia driver to a Brand spankin' new CAT driver.
a b U Graphics card
September 22, 2008 4:07:36 PM

LunaticWolf said:
OH MY GOD!

Look at the Nvidia Driver they are using! 177.41

Thats like 3 MONTHS OLD!


It's still their current WHQL driver on their site, regardless of whether it's 3 months old or not, it's the one nV point you to when you download: Geforce-> Geforce 200 Series-> Windows XP-> English (US);
http://www.nvidia.com/object/winxp_177.41_whql.html

Quote:
It is just a pun on all the people complaining about Toms using CAT 8.6...


That's not a pun, and it's not the equivalent.
People complain about the Cat 8.6 being used for HD4K drivers because they are before the HD4K launch drivers and technical do not include support for the HD4K despite working on the HD4K, and they are also drivers that are older than those that are WHQL certified and on the dowload site.

So your 'pun' is pointless, and as a criticism it's not equivalent even.

Quote:
Aye, but they update far less than Nvidia.


Not for official drivers, they rlease Betas more often, but that isn't even an indication of their updates, and ATi updates official WHQL certified drivers more often and more regularly.
If ATi released all their beta builds who knows which one would have more launches, it would likely see-saw back and forth depending on new architecture and new non-optimized games launching.
Having been involved with the beta programs, there's alot more out there you never see, and that's a difference in their driver strategy. Open versus closed beta programs, there are good and bad aspects to both. But both companies are quick to react to the limitations of their choices.
!