Printable View
Microstuttering doesn't occur in cpu limited scenarios, only deeply GPU limited scenarios(thus why crysis showed it so harshly with the 3870x2's). As such, that's why it didn't show up until 2560x1600, as with the cpu he's using he didn't hit that spot until ultra high resolution.Quote:
Originally Posted by Solus Corvus
In other words, we have not yet seen the end of the Micro-Stutter curse. :(
It's a GRID issue. Only happens in the Menu. But people like Lestat genuinely show that they don't care as long as it gives an opportunity for them to flame.
:down:
Well, It wasn't entirely clear as to what kind of stutter they are referring to. There are also reports of stuttering menus and whatnot. I will agree that we can't consider the case closed based only on Sampsa's tests, more testing needs to be done.Quote:
Originally Posted by DilTech
Do people know the difference between stuttering and micro-stuttering? Because now with the boom of MC they seem to mix the two, and they're very different problems.
ive noticed that playing grid at 8x csaa at 16x12 my vram is almost always at 512MB with rivatuner monitor
cpu never maxes that i have seen. only very occasional stick/lag/...stutter. no not micro stutter obviously.
if 4870X2 gpu temps are now above 90c, what will happen with 2*4870X2?...smoke? :P ...and powersupply requirements for that too?
t800, critical temp in 150 C, don't worry
i doubt that, more like 105-110°C for highend silicon chips, its not some basic transistor we are talking about here.Quote:
Originally Posted by MAS
IIRC the x1950 starts throttling at 115c
Sampsa (@ Muropaketti) has some experience with overheating videocards. :)
IIRC he sometimes forgot to plug the fan after photoshooting for one of his article. I can't find that article anymore but I think the card was 2900XT and temps were way over 100celcius (104C??) when measured afterwards, when the card was already cooled down a bit. Some plastics were melted but card was still fine!
110C yesterday tested with 1.45v...idle 92c
can't control the fan speed.... it was about 14% or whatever default. then i remove the heatsink cover. using 120mm, lower to 39~45 idle and 55 load
I found the article at muropaketti. :)
Quick translation:
http://plaza.fi/s/f/editor/images/0806r600.jpgQuote:
Literally we were horrified in muropaketti test lab, when we "bolted" standard cooler back to HD 2900 XT after LN2 tests. Our misfortune, we connected the fan to the wrong socket of two fan sockets in our engineering sample and booted to windows. Fan wasn't working at all and suddenly we heard big BANG from the red, plastic cover.
In consequence of this we saw smoke arising from the videocard. We quickly shutted down the computer and removed the pci-express sockets from the card. At the same time, when undersigned burned his fingers, we connected a temperature meter between DVI brackets where the card exhausts it's warm air. We measured "unbelievable" 105 degress of celsius from the back of that copper heatsink. When we get our camera ready temperature was dropped whole one degree. We can only guess how high tempereatures went inside the GPU core. All in all, we were lucky and the card is still working fine and this time fan is plugged to the RIGHT socket.
Sampsa Kurri
http://plaza.fi/muropaketti/toimituk...as-koetuksella
Even the best people make mistakes.
I want to see 2x4870x2 temps :)Quote:
Originally Posted by t800
The GTX 280 cards are getting damn hot in Tri-SLI.
http://youtube.com/watch?v=vQdrJCPsWF0
112degrees and one card goes dead.
and thats why god invented water cooling
Well im planning to go 2*4870X2, but based on these temps (even with retail-versions will work cooler) watercooling is solution that i wont need afraid cards will get burned down :PQuote:
Originally Posted by ownage
when are the reviews due?
Mid AugustQuote:
Originally Posted by disruptfam
http://www.pcgameshardware.de/aid,65...k-Test/&page=1
another one.
MS not fixed and still occures when the fps drop in the 4x region. But its a lot harder to pull that off since RV770 is far more powerfull than RV670.
USELESS preview, they used Catalyst 8.5 !!???Quote:
Originally Posted by Hornet331
Much better performance with Cat 8.7. Plus Crysis DX10 1600x1200 full Very High on HD 4870 is much better (avg 24 fps, smoothly playable) than the 1680x1050 (18,7 fps?!) in their chart
Quote:
Diamond 4870x2 Futuremark Testing
http://img255.imageshack.us/img255/3...b59e72dmb0.jpg
Results >>>> http://en.hardspell.com/doc/showcont.asp?news_id=3804
...
regards
if you would have looked at the preview you would have noticed that the use 8.52 (which is beta 8.7, which is more advance then retail 8.7 and every preview that was done till now is with 8.52) and not 8.5. :rofl:Quote:
Originally Posted by Unbornchild
@ Crysis, theres no test in 1600x1200 only 1650x1050, but they test DX10 vs DX9 and X2 fails at the DX10 crysis test, they even stated that in the conclusion, the current beta driver seems to have no DX10 CF profile. :yepp:
Dude you really should chaeck if you need some specs. ;)
Dude they state on their charts they used 8.5.
Indeed i later saw in their system setup they also used Cat 8.7 beta. I already posted that final Cat 8.7 on HD 4870 gives an improvement of 7.5 fps measured in the Crysis Bench Tool (GPU test), over that beta. So, their chartresult of HD4870 is old and horribly outdated. Also, the actual gameplay is accurately reflected in the benchtool Assault_Harbor test, is measured 24 FPS avg in DX10 full Very High 1600x1200 no AA and is absolutely playable.
Their 4870X2 result however can be correct, because the X2 isn't even supported in the final Cat 8.7 yet, so they MUST use that older beta driver. And, so, i expect that a next driver for X2 will give it much better result than what they posted.
So with this new card, it still relies on the driver with a game profile in order for both cores to work?
Yes :(
When there is new game, we have to wait for drivers to get full perfomance.
So I think single gpu is better if you dont like waiting.