Sign in with
Sign up | Sign in
Your question

What's up with the new charts? 280 GTXs sli losing to 8800 GTS sli

Last response: in Graphics & Displays
Share
September 15, 2008 12:03:38 PM

Have I gone crazy or has Tom's hardware lost the way to benchmark????

In COD4 the 8800 GTS sli comes out ontop of every card @ 1280x1024??? Seriously????? When every other benchmark on the internet shows a different story.


The 3870 X2 in CF coming almost at the bottom? I kno its not that powerful of a card, but it can almost match up to 2 8800 Ultras......


Am I wrong? is there something I'm not seeing?

Look at this a 4870 winning against CF 4870s and only gets 80 fps.....while the 8800 Ultra gets 90 and is almost on par

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

September 15, 2008 12:09:34 PM

Yay new charts finally!!!

I dunno, does look odd, maybe just mcuked up some results, odd

Im more interested in seeing how good the 8800GT does in SLi at the lower resolutions, as i have an OC version and was thinking of SLi for my next upgrad einstead of purchasing newer cards. The 8800gt is still a great card.
September 15, 2008 12:13:41 PM

I wouldn't trust those charts, even if they were my last resort. All the results look like the scores were just pulled out of a hat, just like drawing the smallest straw.
Related resources
a b U Graphics card
September 15, 2008 1:02:54 PM

I'm not suprised about the 8800's. More mature drivers will allow them to be on par with other cards at low resolutions. Move the resolutions up a notch, and you see the newer cards power through.

As for the 3870 X2, those are simmilar to the benchmarks that came out when the 8800's were first released. Heck, I remember a few which had the 3870x2 loosing to a 7800GT...(I was buying a comp at the time, so I did a lot of searching for parts).
September 15, 2008 1:13:27 PM

I like the Assassin's Creed bench where single 8800gtx outperforms dual gtx280. :na: 
a b U Graphics card
September 15, 2008 2:15:03 PM

It doesn't surprise me at all if these charts are based off of this article:

http://www.tomshardware.com/reviews/radeon-geforce-comp...

That review had to be the worst review ever posted on Toms. They were using a CPU that was not clocked fast enough to feed the cards, so their charts are all over the place :pfff: . They also used old drivers and a questionable power supply. I would look at the charts from some other sites before making any decisions. There are plenty of other sites who have done valid benchmarks on the new cards.
September 15, 2008 3:21:24 PM

Yeah the scores do seem all over the place. For Quake Wars at 1680x1050 a 320mb 8800gts beats 8800 ultra sli and 3 way sli. . . Other anamoly looking results too.
a b U Graphics card
September 15, 2008 3:33:35 PM

Only good thing I saw was when they said using a 2.93Ghz cpu can cause slow downs
September 15, 2008 3:37:31 PM

gamerk316 said:
I'm not suprised about the 8800's. More mature drivers will allow them to be on par with other cards at low resolutions. Move the resolutions up a notch, and you see the newer cards power through.

As for the 3870 X2, those are simmilar to the benchmarks that came out when the 8800's were first released. Heck, I remember a few which had the 3870x2 loosing to a 7800GT...(I was buying a comp at the time, so I did a lot of searching for parts).


Driver maturity can't justify the low marks gotten by 2 powerful cards, which only 1 280 GTX can almost double teh 8800 GTS:) 

my lord what a bad article....

shortsuff_mt was right on the money with what he said....


Guru3d VGA charts here I come!!!
a b U Graphics card
September 15, 2008 6:30:57 PM

Yea, it's too bad Tom's is having such a hard time putting out a decent review lately. The reviews we end up with are either using outdated hardware, old drivers, or are so old by the time they are posted that they are useless. They have been at least a week behind most of the other main review sites.

If they can't put out a review fast enough for it to be relevant, maybe they shouldn't bother. It just confuses people who still think this is a good place for valuable information. It sure isn't like the Toms of a couple years ago.
September 15, 2008 6:45:27 PM

definitely not The reviews seem to be done by a random selection of people, and the reviews lost their professional touch.

I said this 100 times, but around when the 9800 XT came out Tom's was a big thing for me, now its mainly the forums and community.
a b U Graphics card
September 15, 2008 6:55:22 PM

I agree. I still read some reviews, but by the time they are out I've already read about it in a more professional review somewhere else. I also mainly just come here for the forums. :( 

Have you also noticed what a joke the system builder marathon has become? I just read it to see what weird, outdated, non-overclocking friendly combinations they put together and which crappy tier 4 PSU they used. I hope not to many people base their own builds off of those.

Sorry to take your post off subject. I guess I should start a thread on this topic. I hope I'm not the only one who has noticed it though.
September 15, 2008 9:46:55 PM

naw it doesn't matter that article really pissed me off, what a waste of time (Charts). Your better off not updating at all if your going to put conflicting information.

I'm sorry but that just killed Tom's credibility for me. See when i read Guru3D's (and anandtech, there are more just too lazy to explain em) reviews, they explain the settings well, what test they run, and a personal and professional idea/Input about the item being reviewed
September 15, 2008 10:09:44 PM

I think they would be better off getting rid of the charts and sticking to what they are doing well which is "best cards for the money /month" - which are almost always good, accurate, current articles. Its rare that they are outdated.
September 15, 2008 10:16:01 PM

ovaltineplease said:
I think they would be better off getting rid of the charts and sticking to what they are doing well which is "best cards for the money /month" - which are almost always good, accurate, current articles. Its rare that they are outdated.


Unfortunately that one has been sliding lately too. They somehow decided to put 4870, gtx260, and 9800gx2 on the same $300~ level, and make it sound as if the 9800gx2 perform the same as the other two. I realize that all 3 cost about the same, and had they stuck to the "best bang for the buck" philosophy strictly, 2 very popular cards would be bumped off the article by 9800gx2. But still, at least make a special note explaining the circumstances instead of making it sound misleading. :sarcastic: 
September 15, 2008 10:16:54 PM

ovaltineplease said:
I think they would be better off getting rid of the charts and sticking to what they are doing well which is "best cards for the money /month" - which are almost always good, accurate, current articles. Its rare that they are outdated.


I doubt that even help about 80% of all the readers out there. ATI and NVIDIA pretty much have 1 product at each segment and they are fairly comparible, something which most readers can distinguisher for themselves.

I would perfer that Tom's had a "best cpu" segment instead since the amount of CPUs being thrown to the public from AMD and Intel from month to month is just insane.
a b U Graphics card
September 15, 2008 10:24:08 PM

Looks like an SLi/CF bug, notice the single HD4870 and GTX are at the top.

That to me shows the new cards+drivers having issues with multi-GPU, but the older ones being a little more consistent.

There's a fair amount of anomalies in it, with the HD4850CF being higher than the single HD4850, so you would think that the HD4870CF should be higher than the single HD4870, but overall we're talking about a few frames, that it looks like a driver issue overall.

September 15, 2008 10:29:51 PM

TheGreatGrapeApe said:
Looks like an SLi/CF bug, notice the single HD4870 and GTX are at the top.

That to me shows the new cards+drivers having issues with multi-GPU, but the older ones being a little more consistent.

There's a fair amount of anomalies in it, with the HD4850CF being higher than the single HD4850, so you would think that the HD4870CF should be higher than the single HD4870, but overall we're talking about a few frames, that it looks like a driver issue overall.


It's not limited to sli/cf setups. Stuff like this seems like more than just driver issues... :sarcastic: 
http://www.tomshardware.com/charts/gaming-graphics-char...
http://www.tomshardware.com/charts/gaming-graphics-char...
http://www.tomshardware.com/charts/gaming-graphics-char...
a b U Graphics card
September 15, 2008 10:47:28 PM

L1qu1d said:
I'm sorry but that just killed Tom's credibility for me.


I agree. I've been seeing a downward trend with the reviews lately, but this brings it to a whole new low. They need some decent reviews and they need them fast. This recent trend isn't making me hold my breath though. :pfff: 
a b U Graphics card
September 15, 2008 11:06:23 PM

Looking at the ones you linked to, it looks like a system bottleneck, look at the performance @ 1280x1024 no AA and 1920x1200 w/ 4XAA, very little change, but they do fall into better order once graphics load is increased. In the low res situation like that it would actually still be influenced greatly by drivers and software intereaction at the low end, just like FSX sees wild flips for the same reason (and always has for many sites/benchmarks/reviews), those aren't as surprising.
ET:QW and World in Conflict fall in line when the resolution is increased and AA is added with the stronger cards higher up. ET:QW is still a little off, but then again, 19x12 even with 4XAA isn't all that stressful for the new gen cards, let alone multi-GPU setups.

L1qu1d did link to the higher resolution/settings, but it's still HL2, which is far from stressful even with 4XAA, so system and driver variations still provide a major influence. You're still seeing pretty good return on single cards, so you have less of an issue with the overhead of the CF/SLi profile, with those cards, and then the CF/SLi results become unpredictable.

Only when a single card is already stressed are you certain of seeing a boost in performance from a second card, if a single card is CPU/systen-bound, then adding the overhead of CF/SLi just makes it worse and gives us weird results.

Looks like it's time to retire the old games/tests. Even if you were to make them tougher by adding stuff like texture mods, it lets them show the power of the cards, but it makes the results less reproduceable and relevant to those people who don't use mods or don't even know of such things.
September 15, 2008 11:48:33 PM

My take on this site.

Tom's of the past grew the forum's critical mass. The regular contributors who offer help, insight, lively debate or even entertaining retard fanboism are the real value of this sight. The current owners can focus on global growth while passing off outdated, poorly translated fluff to the core audience.

We the community are the real value of this site. I'm reminded of this whenever I'm googling a hardware question and end up right back here.
September 15, 2008 11:49:43 PM

No1sFanboy said:
My take on this site.

Tom's of the past grew the forum's critical mass. The regular contributors who offer help, insight, lively debate or even entertaining retard fanboism are the real value of this sight. The current owners can focus on global growth while passing off outdated, poorly translated fluff to the core audience.

We the community are the real value of this site. I'm reminded of this whenever I'm googling a hardware question and end up right back here.


Let's start a boycott. :D 
a b U Graphics card
September 16, 2008 12:02:26 AM

No1sFanboy said:
My take on this site...


Yep, agreed.
Some contributors like Don are worth reading, but unfortunately they don't get as much work as they used to.

BTW, when's Top Gear coming Back? Do you know?
I know they said 'Autumn', but we in the colonies call it Fall of course, and here in Calgary, that's like a random weekend at the end of September before we go straight into winter. :sol: 

They're still playing the re-runs of season 10 here (can't believe I pay to watch year-old re-runs). :fou: 
Thank goodness for the .net. [:mousemonkey]

EDIT: HAHa, just looked at your location again... guess you're in the same boat, gotta love BBC-Canada, Eh ! :kaola:  :pt1cable:  :kaola: 
September 16, 2008 12:58:30 AM

Yeah, same boat. Really itching for some new stuff. I've now watched the last season twice, once in pieces on Youtube and again on BBC Canada. Even further off topic did you catch the Arctic special?

Back on topic. I don't like being over critical of some of the guys doing work for Tom's US but in their responses to criticism you kind of get the feeling that there is very little support from the industry and Tom's France. An example was the criticism they took for the case reviews where they defended the piece because these were the cases they had received to review. From that it seems they had no budget to buy any and either the manufacturers don't care or allocate to Tom's in Europe or Asia as a priority. Either way they do a review on products that are a world apart from the stuff that people are asking about in the Home built section or are in peoples profiles.
a b U Graphics card
September 16, 2008 1:19:30 AM

No1sFanboy said:
Yeah, same boat. Really itching for some new stuff. I've now watched the last season twice, once in pieces on Youtube and again on BBC Canada. Even further off topic did you catch the Arctic special?


Yeah that made me laugh, especially since I've been up in the far north (Fort Chimo / Kuujjuaq , QC). Pretty incredible, and that segment was amazing, and almost as funny as the US one. I was also amazed that it was the first time anyone ever tried to drive there? I wonder if that's true, seems weird with the offroad types we have here and in the US.

Just finished the 11th season summer six-pack last night after football, with the funny competition with the German equivalent good end to that set.

Quote:
Back on topic. I don't like being over critical of some of the guys doing work for Tom's US but in their responses to criticism you kind of get the feeling that there is very little support from the industry and Tom's France.


Yeah, it seems that now that it's headquartered in Europe again, it's lost touch with it's US site. Even when the reviews were done by the original crew including Lars/Borsti, the reviews were pretty quickly translated and posted on the English/N.Am. side of the equation. And he also still was able to get the new stuff as quickly as the other major sites. Due to the delay, Tom's seems to get scooped more often than not.

As for the charts, I realize they are tough to maintain, but there almost needs to be a division of charts, because with the inclusion of the X1300 in the same chart as a Tri-G80Ultra or GTX280 SLi / HD4870CF, you kinda lose the usefullness of the two ends because neither is tested in games or settings that are really appropriate for either IMO and they search for a soft middle at best. It's tough though, because some people want to know what the upgrade would be from their GF6600GT to those HD3870 and GF8800GT that are now cheaper than when they bought their GF6600.

To me the charts aren't as big an issue as the other areas, where THG used to excel IMO.
a b U Graphics card
September 16, 2008 1:41:01 AM

You know whos been getting the scoops? Toms far east site. Theyve done well, and I wish theyd do more from there, only IF they wont use NA
September 16, 2008 1:43:19 AM

Sigh....
September 16, 2008 1:51:40 AM

shortstuff_mt said:
It doesn't surprise me at all if these charts are based off of this article:

http://www.tomshardware.com/reviews/radeon-geforce-comp...

That review had to be the worst review ever posted on Toms. They were using a CPU that was not clocked fast enough to feed the cards, so their charts are all over the place :pfff: . They also used old drivers and a questionable power supply. I would look at the charts from some other sites before making any decisions. There are plenty of other sites who have done valid benchmarks on the new cards.


Sigh...Coolermaster's upper PSU's are fine. Real Power is the only PSU line I would buy from them. The rest I wouldn't even consider it. They have used that 850 Real Power PSU on lots of reviews on this site. Research a little.
a b U Graphics card
September 16, 2008 3:33:01 AM

I dont really want to knock my european friends, so dont take it that way, BUT, since theyve gone euro, weve been handed trash. Maybe they need to start working 40hr weeks, and then they wouldnt have started so early, using outdated drivers, non included cards, imature cf/sli drivers showing multi worse than single etc etc. Tsk tsk, shame on Toms for putting this out, as this puts only a larger blackness on the only good eye Toms has left
a b U Graphics card
September 16, 2008 4:13:26 AM

one-shot said:
Sigh...Coolermaster's upper PSU's are fine. Real Power is the only PSU line I would buy from them. The rest I wouldn't even consider it. They have used that 850 Real Power PSU on lots of reviews on this site. Research a little.


Just because Tom's uses the Real Power PSU in it's reviews doesn't make it a good PSU. That is kind of the point of this thread. I don't really have anything against that particular PSU, but is seems like some of the problems they ran into with their SLI/CF setups may possibly be traced to a lack of power as well as their CPU bottleneck. IMO a SLI GTX 280 rig should be packing a little more juice just to be safe.
September 16, 2008 4:14:03 AM

Well, form personal benchmarking, the latest cards show no change in FPS in a change of resolution I only start seeing a difference in 1080p and 1200p.

My old 7950GT gets better FPS at low settings and low resolution than my 9800GX2 or GTX280... So these are for 1680x1050 and above, wouldn't buy them with lower resolutions.
!