the new Radeon 4870x2

[Fodder]

There's also ViewSonic VX2640w 26" which is cheap as well, with 1920x1200 resolution. http://www.digitalversus.com/duels.php?ty=6&ma1=41&mo1=381&p1=3541&ma2=35&mo2=189&p2=1815&ph=1

 

That's a comparison with VX2640w and L245WP-BN 24" (not TN panel)

 

Just thought I'd buy me 2x L246WH-BN as I already have one. It's TN panel but it's pretty close to that 245 I guess. At least closest I could find, hence the comparison :P Anyway, the thing is now to decide whether to go with 3 of the same monitor type or 2 same and then buy a 26" in the middle. Whole day I have been going through different scenarios and more and more variables come across and now when I finally started looking at 40" LCDs I realise I must go to sleep :D

[stealthtemplates]

Will DCS: Black Shark support multi-GPU's? Id also imagine the game will be heavily CPU bound, any news on mutli-CPU also?

[VMFA117_Poko]

AFAIK BlackShark won't support multi-GPU nor multi-CPU. Next modules with new engine will do.

[stealthtemplates]

Damn, guess we're going to have to ramp up the clock speeds on the CPU then if its only going to be single core

[qrazi]

Will DCS: Black Shark support multi-GPU's? Id also imagine the game will be heavily CPU bound, any news on mutli-CPU also?

 

multi-GPU support as in CrossFire(X) or SLI happens on the driver level, not the application level. Therefor DCS:BS doesn't need to support multi-GPU, however AMD/ATi and nVidia should optimize their drivers for DCS:BS.

[Flyby]

Doh! If you're right, qrazi, then I hope ATi and Nvidia are aware of DCS, and Black Shark! Wouldn't that be great?

[Pilotasso]

Damn, guess we're going to have to ramp up the clock speeds on the CPU then if its only going to be single core

 

Not realy. LOMAC is a DirectX8 SIM. DCS will be DirectX9. That alone will bring shader efficiency and eliminates many of the FPS killers in LOMAC. How much that will be ofset by simulation computational demands we still dont know. however some coments have arrisen that if you have a PC that runs LOMAC good now, so will DCS.

 

Recent C2D and C2Q CPU's should run DCS perfectly.

[33rd_bratpfanne.]

sorry, but i have to disagree !

 

LOMAC is highly CPU-Bound. So Quad core does not help you much if you run it at, lets say 2,4 GHZ. With my C2D at this (Stock-)Speed, it was barely faster than my old Athlon 3400+ Single core. As i OC'ed it to 3,2 GHZ, my FPS, especially over Cities, improved significantly. That means more then when changing from an X1950XT AGP to a 8800 GT at the same settings. If DCS is based on the same engine, and with an increased load for "Physics Calculations" on the CPU, be sure that you need a highly clocked CPU more than a actual Graphics card.

 

S~

 

Brati

[Flyby]

interesting point, brat

 

Upon reflection, I tend to agree. LOMAC has always been defined as cpu-bound. IL2 was similarly defined. I suppose FSX is too. Oddly not too many new cpus are burning up the skies with ghz of power, but at least a few can be safely overclocked to 3.6 with air cooling. The e8400 and e8500 come to mind. I don't think LOMAC and Black Shark are so much about how many cpu cores (but at least two, so one core can handle the background stuff) as it is about ghz which I think translates into bandwidth.

Flyby

[qrazi]

Doh! If you're right, qrazi, then I hope ATi and Nvidia are aware of DCS, and Black Shark! Wouldn't that be great?

 

I am pretty sure I am right. However, both CrossFire and SLI don't require driver optimization to work, it's just that in order to get to most out of it, driver optimizations are required. Both companies often get more perfomance on specific games with each new driver release, and that is only part because of bug fixes. They both do a lot of optimalization for popular games.

 

I guess flightsims are way more CPU bound because of the world that is being rendered, and the stuff that is going on in that world.

 

What could be an option is to tap into the graphics card power by using CUDA for calculations such as AI, flightmodel maybe... Ah, I dont know about that, might be completely wrong there.

[Pilotasso]

you guys dont read slow do you? :D

[stealthtemplates]

heres some early preview reviews:

 

http://www.driverheaven.net/reviews.php?reviewid=588

http://enthusiast.hardocp.com/article.html?art=MTUzMSwxLCxoZW50aHVzaWFzdA

http://anandtech.com/video/showdoc.aspx?i=3354

http://techreport.com/articles.x/15105

 

note this is not the final card, but judging by performance its currently owning the GTX280

[Flyby]

thanks for posting the links, stealth

 

I've been leaning towards the 4870 card. But it's really hard to make a decision. Almost too much information. Now add the 4870x2 to the mix. At over $500.00 USD it will outperform the single gpu GTX280, which is just below the X2 in price. Further, it seems to be the reviewers' consensus that ATi needs to really work on the power demands of this card, and the drivers (scaling issues). I don't know about you guys, but power-hungry cards concern me too. The card runs hot as well. I know it's said there will be more optimization, but how much and will it be great enough to really matter? Remains to be see, I know.

But I think what will be best for me is a single GPU solution. I'm not a wealthy man, so it's really a no-brain-er. I'll need to see a comparison between the 4870 and the '280 in FSX to see how they handle the busy world of a modern flight sim. Then I'll hope the price of the '280 drops again! :D Imagine if that happens? I've seen where the 280 in SLi outperforms the X2 in Crossfire. Prices will drop again one day making the purchase of a second card a nice upgrade down the road.

What a headache!:doh:

Flyby out

[stealthtemplates]

Just get the HD4850/HD4870, they cant be beaten on price / performance ratio.

 

Would you really spend twice the amount to get only 10% more performance by buying a GTX280?

 

The drivers are very early in beta stage and its not even the final BIOS for the HD4870X2 so I wouldnt judge the performance of the card based on those reviews, it will be better. You can also easily overclock the HD4870 to get the same performance of the GTX280 ;)

 

FSX is terribly optimised so I wouldnt base a card purchase just on that, that would be kind of stupid considering how highly priced they are.

 

The GTX280 has already dropped $150 and its still not competitive, dont ever think it will go below $400 anytime soon.

 

If you havnt got much money dont go and waste $449 on a GTX280 when you can get very close performance for $150 less :)

[Flyby]

actually,

 

My most solid plan was to buy the 4870 card. I hope it is easily overclocked as that would be a bonus. I figured I'd need at least an X38 chipset to eventually run two cards in Crossfire, both at x16. The price of a 4870 will drop again one day, hopefully just in case I want to go that route. So the savings in selecting the 4870 could be put towards the mobo. I noticed that a lot of the tests use a quad 9xxx cpu clocked at about 3.6 ghz. I'm not sure why the use of a quad cpu for testing. Any ideas? I wonder why not test with, say, an E8500 clocked to 3.8 ghz? Maybe it's because the games tested cam make use of multiple threads?

I mention FSX because it's the only flight sim I've seen in testing lately as there are no newer releases about. Actually a new combat flight sim is what I'd like to see in testing. But it will probably be more cpu-bound that gpu-bound. I tihnk any of the newer cards would be fine for that kind of simming, 4850, 4870, GTX260, etc.

Flyby

[Kuky]

if anyone wants to run HD4870 in CF or get HD4870X2 make sure you have 850W good quality PSU as from personal experience I have found my dual link 650W+550W PSU does not quite live up for oc CPU and so much power required for high end multi GPU setup. Every one of my voltages are but under optumum and +12V is actually showing in BIOS hardware monitor as 11.87V and even dips to 11.81V just if fan on HD4870 spins up fast... obvious sign that this PSU does not cut it for this setup, I even get system restars just like that... very random... and I was wondering what was causing it, untill I checked in BIOS and saw all my voltages are slightly bellow...

[stealthtemplates]

They use a quad-core to eliminate any CPU bottleneck that may be present, some games like Crysis take advantage of quad-core so I guess thats why they test it.

 

What brand is your PSU? From what ive seen a 620 watt PSU is fine to run, as long as its good quality. Never risk powerful hardware to a generic / sub-generic PSU's.

Edited July 15, 2008 by stealthtemplates

[G3]

some gays like Crysis

 

com'n leave the gays out of this :)

 

serious now:

as for power supplies i totally agree, why skimp on psu, its most critical for stability.

also i hear that thanks to this new physix or cuda business you can now use the gpu for folding@home research etc, nice, just like another cpu core, so they say.

Edited July 15, 2008 by G3

[stealthtemplates]

didnt realise i wrote that :P

 

its not physx or cude that does folding at home, its in the program itself.

very interesting to see these cards optimised for folding@home, with all the terraflop power its a lot more powerful then a PS3 could ever pull off

[Flyby]

They use a quad-core to eliminate any CPU bottleneck that may be present, some games like Crysis take advantage of quad-core so I guess thats why they test it.

 

I wonder how many of those tested games actually take advantage of quad cores? I wonder if suing a quad core is overkill. I will have to write to some of these test sties and ask why they use quad cores with large caches instead of perhaps a q6600? Would even a dual core like the E8500 do? It's supposed to be easily overclockable to nearly 4ghz with standard cpu cooler. Of course I'm assuming it's about bandwidth and ghz.

My point is it seems odd to me to read a review where the object is try to get as much from a gpu test as they can, only to present results that the average pc gamer (system) might not achieve. Why not test to the average system? It might reflect real world results, and might even make decision-making a bit simpler. As it is now, the test results seem to require a "filtering down" in order to guesstimate real world results.

Whew! it got windy in here!:megalol: sorry.

Flyby out

[stealthtemplates]

Also need to take into account that they use programs such as 3dmark when benching, to get a good score on that a quad-core is essential

[Flyby]

So in some instances a sufficiently overclocked quad is better suited for testing a gpu than a similarly clocked dual core processor? I can see this where the particular test can take advantage of multi cores. It makes sense in that vein. I suppose that's it then. 3.6ghz on a single core does not perform the same as 3.6ghz on a dual core which does not perform the same as a 3.6ghz on a quad. Must have something to d with all those little Windows processes running in the background, sucking up resources.

[Flyby]

link to a good article on multicore processors/gaming

 

here's a good read on multicore processors. I'm sure most f you have read it.

http://www.guru3d.com/article/cpu-scaling-in-games-with-quad-core-processors/1

 

 

Tell me what you think.

Flyby out

[Pilotasso]

These benchmarks have a major flaw, they are all but one of them from first person shootemups. The biggest factor in them is how much data can be fed to the graphics card. Computational demands on the CPU is second nature.

 

this makes all CPU's including AMD's offering look strikingly similar to each others. Not everybody is a heavy overclocker, though the duals sure offer more headroom.

 

Having said that I would love dual core users play Supreme commander at the settings I am used to. :) In all other games there will be no difference between dual or quad of same speed. However in video encoding I give blessing to the ammount of videos I can code at any time. :)

Edited July 15, 2008 by Pilotasso

[Flyby]

Pilotasso, need a little interpretation

 

what settings are you used to playing SC at? I am not familiar with that game, but I think you imply that where there is a heavy cpu load a quad core is better than a dual core at the same ghz. So maybe a combat flight sim, survey style, would be a better indicator? Do you tink a study combat flight sim would be better run on your q6600 at 2.66ghz versus an E8400 at 3.0ghz? Maybe it depends on how many cores a sim can make use of? You see, I'm looking for the definitive approach to building a system for combat flight sims only, since that's all I'd play.

Flyby out