I'm a bit in the same situation and decided to stay on my current 580GTX and not mgrate to the Keplers. Even worse I didn't looked right and now have a 2,5slot design in my rig which blocks the other two free GPU slots.....
Why don't you guys just not keep your current GPUs and then just mount in other ones if you have a more clear image of what you need? GPU are the most expensive part of a computer but can be switched easily. I think you boss would understand because you show responsibility for his money doing so and not just telling him "hey boss, I have no idea how market will develop in next 6-8 months but lets buy 4 GPU and see how it goes."
tyan + k20
PURE3D Visualisierungen
Sys: Intel Core i9-12900K, 128GB RAM, 2x 4090 RTX, Windows 11 Pro x64, 3ds Max 2024.2
Sys: Intel Core i9-12900K, 128GB RAM, 2x 4090 RTX, Windows 11 Pro x64, 3ds Max 2024.2
Thanks for all the information here! I was waiting for some news 'bout these new units.
Kinda happy to hear this team is looking forward to stick to mainstream cards rather than Pro units =)
One the other hand it would be nice to have these new expensive toys benchmarked at least =)
As for some GTX cards might be not be an option, because of lower available vRam
(on Fermi we had 3gb GTX while 6gb is available onTesla/Quadro line)
.. knowing nVidia, I doubt they are going to change their product strategy for big Qs & Ts vs. GTX =)
Kinda happy to hear this team is looking forward to stick to mainstream cards rather than Pro units =)
One the other hand it would be nice to have these new expensive toys benchmarked at least =)
As for some GTX cards might be not be an option, because of lower available vRam
(on Fermi we had 3gb GTX while 6gb is available onTesla/Quadro line)
.. knowing nVidia, I doubt they are going to change their product strategy for big Qs & Ts vs. GTX =)
Ghm, Water Cooling might be an option =).. Even if it has it's price, but benefits for compactness, noise & stability =)mbetke wrote:I'm a bit in the same situation and decided to stay on my current 580GTX and not mgrate to the Keplers. Even worse I didn't looked right and now have a 2,5slot design in my rig which blocks the other two free GPU slots....."
-
gabrielefx
- Posts: 1701
- Joined: Wed Sep 28, 2011 2:00 pm
http://randomcontrol.com/blog/?p=867
someone did different tests with Arion. I think they are good.
Vlado from Chaos reported that Octane was slight faster with the K20x. He optimized the code of Vray-RT to get better performances from the K20.
Probably Otoy should buy one K20 to check if the kernel needs some tweaks.
I was on the point to buy 3 K20s, but after reading the tests performed by Chaos I stopped my purchase.
Now I need to know if Octane runs well on K20s.
someone did different tests with Arion. I think they are good.
Vlado from Chaos reported that Octane was slight faster with the K20x. He optimized the code of Vray-RT to get better performances from the K20.
Probably Otoy should buy one K20 to check if the kernel needs some tweaks.
I was on the point to buy 3 K20s, but after reading the tests performed by Chaos I stopped my purchase.
Now I need to know if Octane runs well on K20s.
quad Titan Kepler 6GB + quad Titan X Pascal 12GB + quad GTX1080 8GB + dual GTX1080Ti 11GB
Wow Gabrele, thanks for the link! Very interesting and tbf, the first actual "benchmark" of the new Tesla i have seen (and probably the first out there).
Anyway, i dont get the numbers, granted, i was never good at math
But the guy says, its 261 for Tesla and 163 for GTX580 - apparently that means Tesla is 50 percent faster... since when 2x163 = 261? My calculator says its actually 38 percent faster, hehe.
Additionally, gtx580 is 10 percent faster than 680. But you can see there 155 next to gtx680 (though there are more of them with lower numbers)...still, i am pretty sure difference between 155 and 163 is not 10 percent...
Bottom line anyway - tesla with 2496 cores = 261
gtx 680 with 1536 cores with = 155
1536/2496 = 0,615
155/261 = 0,59
disregarding diffferent clockspeeds, its scales almost perfectly core-wise, therefore it seems, there is nothing special about the tesla beyond bigger number of cores. Gimme dual GPU geforce with 2 of these tesla chips for about 600/700 Euros as i paid for 590 and personally, i´ll be happy camper.
Anyway, i dont get the numbers, granted, i was never good at math
But the guy says, its 261 for Tesla and 163 for GTX580 - apparently that means Tesla is 50 percent faster... since when 2x163 = 261? My calculator says its actually 38 percent faster, hehe.
Additionally, gtx580 is 10 percent faster than 680. But you can see there 155 next to gtx680 (though there are more of them with lower numbers)...still, i am pretty sure difference between 155 and 163 is not 10 percent...
Bottom line anyway - tesla with 2496 cores = 261
gtx 680 with 1536 cores with = 155
1536/2496 = 0,615
155/261 = 0,59
disregarding diffferent clockspeeds, its scales almost perfectly core-wise, therefore it seems, there is nothing special about the tesla beyond bigger number of cores. Gimme dual GPU geforce with 2 of these tesla chips for about 600/700 Euros as i paid for 590 and personally, i´ll be happy camper.
Intel Core i7 980x @ 3,78GHz - Gigabyte X58A UD7 rev 1.0 - 24GB DDR3 RAM - Gainward GTX590 3GB @ 700/1400/3900 Mhz- 2x Intel X25-M G2 80GB SSD - WD Caviar Black 2TB - WD Caviar Green 2TB - Fractal Design Define R2 - Win7 64bit - Octane 2.57
this would have been a nice alternative
http://www.pcper.com/news/Graphics-Card ... aunch-Soon - let's call it 690 with 8GB of vRAM =p though we probably will not going tosee one at all =)..
http://www.pcper.com/news/Graphics-Card ... aunch-Soon - let's call it 690 with 8GB of vRAM =p though we probably will not going tosee one at all =)..