HardOCP's SLI Upgrade Guide

Just what I needed to hear. Thanks Kyle and Brent for the article. It seems that it would be worthwhile to wait a couple of months before purchasing any Ultra video cards, so they can fix that zero frames problem in HL2 if the manufacturers were right in saying it was a BIOS problem. BTW if the CPU is a major factor in these new games, how can people with A64 3000+ expect their results to be?
 
Definetly a good read, I wish I had enough to loot to upgrade. Nvidia should throw together SLI Package Deals where they sell an Nforce 4 motherboard and two SLI-ready video cards for a cheaper price. They could sell that Asus board and two 6600GTs for like $650
 
great read.

thnx for the review guys. great job on it.

i myself can't wait to get it. 6800GT's here i come....... :D
 
Good review as usual.

One thing I wonder about Doom3 benchmarking in general, would it give a more suitable result if you ran with com_fixedtic -1 in the console to uncap the fps? The HardOCP reviews I've read always comment on the 60fps cap.

Obviously fps peaks much past 90 aren't a great deal of use in gameplay, but it'd be interesting to see what happens when Doom is uncapped.
 
mulpsmebeauty said:
Good review as usual.

One thing I wonder about Doom3 benchmarking in general, would it give a more suitable result if you ran with com_fixedtic -1 in the console to uncap the fps? The HardOCP reviews I've read always comment on the 60fps cap.

Obviously fps peaks much past 90 aren't a great deal of use in gameplay, but it'd be interesting to see what happens when Doom is uncapped.
Click to expand...
We can turn on and off all sorts of things in games when we evaluate hardware. The point is to show how a realworld gaming experience will be impacted and we think the overwhelming majority of our readers will not be using the com_fixedtic -1 console command line.
 
Maybe I am missing something, but from the benchmarks, the SLI config really didnt make much difference at all for the 6800 ultra's. Looking at HL2, all it got was about 10 fps more on average than the xt-pe, but it had a min of 3 !! I don't know, but they just don't seem comprable to the initial benchmarks we have been shown. Though it did seem to shine in Doom 3, making UQ at full settings a reality :)
 
Very interesting article indeed. Just out of academic interest; did you run 3DMark benchmarks or similar with some of these setups (e.g. the 2x 6800Ultras SLI)...?

Any visionaries out there seeing the future of SLI technology? Maybe in 2-3 years from now we will be upgrading video cards like we upgrade memory today, and no decent motherboard will have less than 4 GPU sockets... with integrated snap-on cryo-fluid coolers...:)
 
swflbatth said:
Maybe I am missing something, but from the benchmarks, the SLI config really didnt make much difference at all for the 6800 ultra's. Looking at HL2, all it got was about 10 fps more on average than the xt-pe, but it had a min of 3 !! I don't know, but they just don't seem comprable to the initial benchmarks we have been shown. Though it did seem to shine in Doom 3, making UQ at full settings a reality :)
Click to expand...
As mentioned, in HL2 you are dealing with CPU bottlenecks and such in realworld gameplay. Also we did fully discuss the pauses and hitches in the article.

Our point is that this is REAL GAMEPLAY and not a benchmark from a timedemo. We stopped using 100% timedemos about 2 years ago because they do not show realworld game experience issues. Quite frankly, I would like to not use them at all, but sometimes it is demanded, by our readers.

I think this is a great followup to the HL2 benchmarks article as it shows the real deal.
 
jon67 said:
Very interesting article indeed. Just out of academic interest; did you run 3DMark benchmarks or similar with some of these setups (e.g. the 2x 6800Ultras SLI)...?

Any visionaries out there seeing the future of SLI technology? Maybe in 2-3 years from now we will be upgrading video cards like we upgrade memory today, and no decent motherboard will have less than 4 GPU sockets... with integrated snap-on cryo-fluid coolers...:)
Click to expand...
No, we did not run 3DMark, but in a couple of weeks it should be hard to find many review sites that do. We play games not benchmarks. ;)
 
Remember the Futuremark scandal with [H]ardOCP about bashing their 3dmark lineup. This is why we don't expect to see any numbers from [H]
 
I understand the reasons behind the benchmarks, but they did seem weird.
Since I did my own HL2 benchmarks a couple days ago using your demos, 1600x1200x32 4xAA/8xAF is the standard I'm looking at. In the Canals time demo, I got 96 FPS on my single Ultra OC (sig below), my card OCed to 451/1200. Your result with the Ultra SLI setup was 94.2 with the 16xAF and reflect water to all. I'll have to go home and run the timedemo at those settings and see what I get, your number is 66. Yes, I'm looking for an apples to apples comparison from the previous HL2 article since I ran the benchmarks myself. Like I said, I fully understand the reasons, but when I'm getting 96 fps in a benchmark, and then yours comes out at 94.2 with detail cranked up, it means I've won. I dunno, hard to describe that, I'm not going to play at 16xAF, 8xAF is fine. I was looking to see a number like 150 fps or something at 4xAA/8XAF, but it's not happening.

While SLI is great for those wanting to buy another card in a year, I wouldn't get too excited about it. In the time of a year, Nvidia will come out with PS4.0, or some new rendering technology. Those who have the SLI setup will get fast frame rates, but they won't see the eye candy a new technology brings.

I think the wait will be 3 months before we see this on the street. With black Friday and Christmas coming, I'd buy now and not fret over waiting. A 6800 Ultra OC paired up with an FX55 screams thru all these games right now, and the hardware is available.
 
Xilikon said:
Remember the Futuremark scandal with [H]ardOCP about bashing their 3dmark lineup. This is why we don't expect to see any numbers from [H]
Click to expand...

Oops, I didn't know that... sorry
 
Hondaman4ever said:
Just what I needed to hear. Thanks Kyle and Brent for the article. It seems that it would be worthwhile to wait a couple of months before purchasing any Ultra video cards, so they can fix that zero frames problem in HL2 if the manufacturers were right in saying it was a BIOS problem. BTW if the CPU is a major factor in these new games, how can people with A64 3000+ expect their results to be?
Click to expand...

The 0FPS problem in HL2 wasn't an issue with the BIOS, that seems to be a game issue, it happens on ATI and NVIDIA hardware and others are reporting it happens with HL Source and Vampire Bloodlines, so it sounds more like a Source engine problem to me.
 
Lord of Shadows said:
How hot were these things? Were any special tricks needed when overclocking in sli?
Click to expand...

they did not get hotter than usual, i don't really see heat as an issue, with this motherboard there is plenty of room between the double slot cards

no special tricks needed, when you OC it just shows you one OC menu as usual, but it OCs both cards, so basically you are bound by the lowest common denominator

for example if card A can OC stable at 450/1200 but card B can only do 440/1200 then your OC will be 440/1200
 
jon67 said:
Come on, don't tell me you're not even a little bit curious about what the dual 6800U's can do with the Proxycon scene...
Click to expand...

i didn't even think to run 3dmark, heh

i was concentrating on games ;)
 
Xilikon said:
Remember the Futuremark scandal with [H]ardOCP about bashing their 3dmark lineup. This is why we don't expect to see any numbers from [H]
Click to expand...

more like, we just don't see any value in it for gamers

gamers comes from the root word, game ;)
 
Great article guys. I really like seeing the benefit of running the SLI over just one card. Your method of getting the best playable settings really stands out in this guide.
 
NoGodForMe said:
I understand the reasons behind the benchmarks, but they did seem weird.
Since I did my own HL2 benchmarks a couple days ago using your demos, 1600x1200x32 4xAA/8xAF is the standard I'm looking at. In the Canals time demo, I got 96 FPS on my single Ultra OC (sig below), my card OCed to 451/1200. Your result with the Ultra SLI setup was 94.2 with the 16xAF and reflect water to all. I'll have to go home and run the timedemo at those settings and see what I get, your number is 66. Yes, I'm looking for an apples to apples comparison from the previous HL2 article since I ran the benchmarks myself. Like I said, I fully understand the reasons, but when I'm getting 96 fps in a benchmark, and then yours comes out at 94.2 with detail cranked up, it means I've won. I dunno, hard to describe that, I'm not going to play at 16xAF, 8xAF is fine. I was looking to see a number like 150 fps or something at 4xAA/8XAF, but it's not happening.
Click to expand...

you are running timedemos

we are running real world gameplay run-through's in realtime

and I think you've just realized yourself how they differ ;)

so now, figure out which one you'd like to believe in

timedemos, or actually getting in the game and playing it
 
botreaper10 said:
Great article guys. I really like seeing the benefit of running the SLI over just one card. Your method of getting the best playable settings really stands out in this guide.
Click to expand...

Thanks!
 
Only thing i want to know that would bring real value to an SLI setup is can you run Quad monitors? Think of the fun you can have with like just three monitors for like CS! Front, left and right!! bwahahahah!!


Would that be cheating?? hehe

-P
 
Pharacon said:
Only thing i want to know that would bring real value to an SLI setup is can you run Quad monitors? Think of the fun you can have with like just three monitors for like CS! Front, left and right!! bwahahahah!!


Would that be cheating?? hehe

-P
Click to expand...


yes i would like to know to. can you run 3 monitors?
or even 4?

plz share :D
 
In the December issue of Maximum PC they deal with SLI and confirm that you can indeed run 4 monitors at once.

Now all i need is a boat load of monet to upgrade my rig!!!
 
With the matrox parhelia, they had to have special circuits on the board as well as drivers for a game to run across 3 monitors. When it was first released, I thought the parhelia was the greatest idea ever, but with the decline in large lcd prices and widescreen finally becoming the standard, I don't think there is a big market for multiple screen gaming.
 
Well, for the general case, I guess I could say I am disappointed. Now I know things will only improve with future drivers and such, but SLI only allowed nvidia to go from 2xaa to 4xaa on farcry. I mean for $500 more, I think I would cry. Currently, it isn't the performance gain they were making it out to be (except on Doom 3), or maybe it just wasn't the performance gain I was looking for. Perhaps things will get better when you all receive the new pair of ultra's.
 
[RIP]Zeus said:
yes i would like to know to. can you run 3 monitors?
or even 4?

plz share :D
Click to expand...

i think just 2, i haven't tried it though, don't have 4 monitors to try it out on
 
Out of curiosity, why do you benchmark Doom with the 60 fps cap still enabled? It's really simple to disable it: seta com_fixedtic -1 in the console. I think this would give much more useful results for seeing how things stack up--some of us still don't think 60 fps is good enough.
 
Suicider said:
Out of curiosity, why do you benchmark Doom with the 60 fps cap still enabled? It's really simple to disable it: seta com_fixedtic -1 in the console. I think this would give much more useful results for seeing how things stack up--some of us still don't think 60 fps is good enough.
Click to expand...

If you don't mind, could you perhaps elaborate and explain to me why you think 60FPS isn't enough?
 
Brent_Justice said:
i think just 2, i haven't tried it though, don't have 4 monitors to try it out on
Click to expand...

What about some screenshots with like 8xSAA / 16xAF at 1600x1200
or some other unplay/runable setting on anything but the 6800Ultra SLI.

Just so I can see how pretty... :)
 
Brent_Justice said:
If you don't mind, could you perhaps elaborate and explain to me why you think 60FPS isn't enough?
Click to expand...

I notice stuttering unless I have 100hz on the moniter... and 100+ fps
But thats me... I have very "sensitive" eyes.


in an intensive fire fight with grenades going off everywhere, etc... it can lag you enough where you miss one getting off one shot gun blast to kill the enemy.
 
Brent_Justice said:
If you don't mind, could you perhaps elaborate and explain to me why you think 60FPS isn't enough?
Click to expand...

Mainly because I can tell the difference as the fps goes up. It feels better and looks better to me. A big part of it is the refresh rate: if I'm at a constant 60 fps but 100 hz refresh, it looks jerky and kind of bugs my eyes. 60 fps is fine for gameplay, but obviously the hardware can and will do better--my single 6800U can do better. So if the hardware can handle better, why settle for 60?
 
Brent_Justice said:
you are running timedemos

we are running real world gameplay run-through's in realtime

and I think you've just realized yourself how they differ ;)

so now, figure out which one you'd like to believe in

timedemos, or actually getting in the game and playing it
Click to expand...

I hope they fix that bug, because fans want to benchmark their card along with your results. It's kinda hard to do that if I have to run thru an entire level. I'm glad you published your timedemos from the previous article. Now I can go home and run my card cranked up to the max and see how it compares.
 
Suicider said:
Mainly because I can tell the difference as the fps goes up. It feels better and looks better to me. A big part of it is the refresh rate: if I'm at a constant 60 fps but 100 hz refresh, it looks jerky and kind of bugs my eyes. 60 fps is fine for gameplay, but obviously the hardware can and will do better--my single 6800U can do better. So if the hardware can handle better, why settle for 60?
Click to expand...
Indeed, if the fps match the refresh rate of the monitor, it is as best as you can go. I feel it too with my eyes.

However, Brent is a little right too on it because the eye cannot really discern 60 fps and 80 fps.
 
Xilikon said:
However, Brent is a little right too on it because the eye cannot really discern 60 fps and 80 fps.
Click to expand...


Im waiting on the inevitable "omg i can tell the difference between 90fps and 95fps" post. It always happens someone whos eyes just happen to be better than everyone elses.
 
Xilikon said:
Indeed, if the fps match the refresh rate of the monitor, it is as best as you can go. I feel it too with my eyes.

However, Brent is a little right too on it because the eye cannot really discern 60 fps and 80 fps.
Click to expand...

Oh let's not get into that again.

BTW IMO, the eye can see the change in FPS. Each person does percieve things differently though.
 
I can indeed see framerate chages. One of the reasons why I always try to have the fastest hardware.

:cool:

I also can see up to 100 fps.
 
Looks great guys... I guess it's articles like this that kill roundups :D

Anyway, I know this article is to feature the video card tech but will we see a review of the A8N-SLI soon?

Also, you guys are pretty good about this but I think this one will deserve a few updates as SLi matures.

Great read.
 
You must log in or register to reply here.
Back
Top