Nvidia Quadro K2000M

Well there was no support for FP64 in the old Quadro 2000M, so no change there... But it is really sad that everybody is just guessing and there is no real information available. I am deciding if I should go for old W520 (I prefer original keyboard better) before it is discontinued or for W530, but I really need to know something about its GPU performance. I plan to use it for some bioinformatics software with CUDA support and all reports show that Kepler based GPUs are not that great for GPGPU so far.

 

Is there a way to find out what Chip its actually using(Fermi or Kepler) When you have a unit in front of you? I have a W530 on the way with the K2000M.

also, If this hasn't been posted

http://www.nvidia.com/content/PDF/features-benifits/Quadro_mobile_features_benefits_final.pdf

Not very helpful though....

 

obviously they are using kepler, the k5000m has 1344 cuda cores(same as gtx 680m), thats more cores than there have ever been in a fermi card, about 2 and half times more, not to mention its mobile so it must be kepler, you guys are so ignorant.

also in the link from the last post above me, it says pcie 3.0, only kepler has pcie 3.0 duh.

edit: and the new kepler desktop quadro cards are called the k10, and k20, hmm i wonder what the k stands for?????

 

Known applications for sys-info:
GPU-Z[ but the support for mobile Kepler is still buggy ]
HWinfo32 or HWinfo64 depending on your operating system[but I guess now everything comes with Win64]
Aida64

Unfortunately there is still no way to tell if it is a GK107 or something that still has 384 Kepler-shaders, but has more GPGPU Power and not due to optimized drivers.

Could you also do some computing benchmarks? We are all curios about how it performs. Here is a list:
1. SpecviewPerf11
2. Luxmark
3. ComputeMark
4. GPU Caps Viewer[there are a few benchmarks in the drop-down list]
5. TessMark[also included in GPU Caps Viewer]

 

Code:

Unless someone beats me to the punch, I'll run the benchmarks once I get that machine.

 

Yet they are very hesitant in releasing Kepler for Desktop Quadros because of the lack of FP64... Then again, is there any information on the K5000M? I have heard nothing beyond K2000M...

 

here: http://www.nvidia.com/content/PDF/features-benifits/Quadro_mobile_features_benefits_final.pdf

and here: http://www.nvidia.com/content/PDF/product-comparison/Quadro_Mobile_Product_Comparison.pdf

the k series of quadro are all kepler. thats it, i dont even understand why there is doubt, if the card had 384m fermi cores it would be a 100w card, i dont think they would put that in a lenovo, and especially not at that price point.

 

The product comparison table says the K2000M is PCIe 2.0. the nVidia control panel on my system also shows PCIe Gen 2

 

Hi Fullmetal99012

Can you post a GPU-Z screen-shot please
Has the notebook optimus?

Thx

 

NP. Here it is showing PCIe 2.0

Yes, The machine has optimus (ThinkPad W530)

EDIT: nVidia Still doens't have Verde Drivers supporting this card yet, so maybe that will fix the GPU-Z not showing physX or DirectCompute.

 

It appears I was wrong.

The K2000M does have similar GPGPU power with the 2000M and it is based on the GK107 with no extra computing units stuffed in it. For now, we've got the ScpeViewPerf11 Benchmarks over at Notebookcheck.com -> Thinkpad W530 Review.

Let me see if I can fiddle around with the differences[comparison with W520, the Dell seems to score a tad higher in some tests, maybe it is higher clocked?]:
In Catia, K2000m wins by 19%
In Ensight, 2%
In Lightwave, 11%
In Maya, 4%
In Pro/Engineer, 37%
In SolidWorks, 8%
In Tcvis, 24%
In Siemens NX, 22%

Average improvement over the 2000M: 16%. Now, what if you clock it from 550Mhz to 745Mhz like the Kepler is? I honestly do think the only thing that makes the Kepler win this fight is the clock speed. Needs further investigation.

Which leads us to: Fullmetal Can you kindly provide some HWinfo64 screen-shots? And also, your results on the list of benchmarks I provided?

 

I guess I was right the whole time... It is still a Fermi.

 

Are you being ironic about the performance? I just said it is a Kepler, GK107.

 

Yeah, that confused me too.

 

Didn't read right. Nevermind...

 

Ill post HWInfo64 once I get it working. It doens't seem to detect my nvidia card like GPU-Z does.

 

It is probably not updated yet. The Quadros are often a bit slow in the updates to both GPU-Z and HWiNFO when compared to the GeForces.

 

My bad, I thought I read in one of the Hwinfo updates that they've added support to the new Kepler Quadro. Hwinfo were the first to report correctly the GK107 from GTX 660M so maybe I extrapolated too quickly that they managed to implement the K2000M. We'll just have to wait a bit more

@Star Forge: No biggie

 

Hi guys, need your help.

I will be using the following software:

-Maya
-Zbrush/Mudbox
-Houdini
-Alias automotive
-Adobe CS5
-V-ray

Will the k2000m be a card that makes good use of this software? Or will it mostly be the CPU?

If not, any other suggestions on a gpu (cosumer or professional).

Thank you in advance.

 

I use Maya/3DSMax/Autoworks/Solidworks and they all use the K2000M. I know my CS5.5 Works with the K2000M. as for the others, If they support GPU Rendering, they will support the K2000M, I even use MATLAB, and that supports the CUDA on the card.

(If you get a machine with optimus, you may have to set it to choose the Quadro for these programs).

 

I thought I read there was a way in the bios when booting to set everything to the dGPU and vice versa with the iGPU. Is that true?

Sent from my Galaxy Nexus using Tapatalk 2

 

on some machines yes. others lock you into optimus. Also, I like using optimus, it seems to save battery power(when it works properly).

 

i've just recieved my Thinkpad with k2000m.

What is the best driver and where can i get it?

 

From Lenovo. NVidia hasn't released proper drivers for windows yet, or they aren't available on their website.