Passmark Benchmarking on virtual desktops

[Levsen]

Hi,

Iīm new to passmark. I know it is possible to run the passmark test on virtual desktop. My questions is how meaningful the results would be.

I have some people who are complaning that their virtual desktop is really slow, so we are building a new environment. So if i only run one machine at a time and perform the passmark test, would the results be meaningful, that for instance our new environment is faster? Or would this be not reliable at all?

Also another questions (not related), can someone give an overview of the passmark overall scoring %age? Something like Total score = 40% CPU, 20% memory etc. etc..

Sorry for this (noob/weird) questions, happy about all answers!

[David (PassMark)]

Should be no problem doing this and getting meaningful results.

You need to be aware of what you are doing however and apply some common sense. For example running a 3D test on a web server doesn't make much sense.

The exact formula for the overall PassMark rating can be found here,
Yes, the CPU gets the most weight. Then disk and 3D. Then 2D & RAM.

[Levsen]

Should be no problem doing this and getting meaningful results.

You need to be aware of what you are doing however and apply some common sense. For example running a 3D test on a web server doesn't make much sense.

The exact formula for the overall PassMark rating can be found here,
Yes, the CPU get the most weight. Then disk and 3D. Then 2D & RAM.

Hello David,

first of thank you so much for the reply and the valuable information!

And yes I agree with you, after doing some test the 3d test scores are obviously very low (they vary from 10-15 score).

My observation is that the overall passmark score on VDs are between 100-200, and on my normal laptop they are about 1k-1.5k. The problem i see is now, because of the low "3d graphics mark" score, my overall benchmark score is really low on VD systems, compared to normal PCs/Laptops.

Is there any meaningful way in applying my own score which excludes 3d test? So I can have reliable and comparable scores through the platforms?

The VD almost performs better in every aspect vs my Laptop, except for the 3d Graphics score, but still scores 10x lower than my laptop (Instance: Laptop PassMark Rating: 1371 vs VD PassMark Rating: 147.6). Since I focus on the working aspects of computers/vd, I would love to find a solution to get a reliable/comparable score between the platforms.


I donīt know if this will help make things clear, but i will post the scores of the 2 runs (My Laptop vs my VD).


Virtual Desktop:

PassMark Rating:
147.6

CPU Mark:
2699 (2084/3248/4.68/14.6/4004/580/224.5/2453/172

2D Graphics Mark:
515 (26.4/114.6/275.7/115.5/538/372/8.3)

3D Graphics Mark:
12.5(1.46/0.54/-/-/-)

Memory Mark:
1323 (58/12014/7637/7584/2176/36.4/14831)

Disk Mark:
5622 (144.4/588/822/-(CD/DVD Read))


Laptop:

PassMark Rating:
1371

CPU Mark:
3296 (5524/2374/9.7/14/4184/565/188.2/2337/136

2D Graphics Mark:
428.9 (19.0/76.6/111.0/76.1/444.3/560/10.4)

3D Graphics Mark:
571(42.8/25.5/9.8/6.6/227.0)

Memory Mark:
1348 (55.4/19157/8693/4294/3354/36.9/8895)
Disk Mark:
581 (88.1/70.9/1.64/-(CD/DVD Read))
(Numbers in brackets are the sub test scores of that category in order (For instance: CPU Integer Math; CPU Floating Point Math; CPU Prime Numbers etc.))

Thanks for taking time for my problem!

Cheers,
Alex

[David (PassMark)]

Well yes, the 3D performance on the VD is truly abysmal compared to your laptop. And your laptop itself has 3D scores that are really poor. If you are ever going to do any 3D work on these VDs, then abandon that thought now. This has the additional downside, in that some applications are now starting to use the 3D hardware for compute tasks (called Direct compute and OpenCL). This can speed up processing for non 3D tasks. But your VD doesn't seem to run Direct compute at all.

Anyway assuming you don't care about 3D, the simple solution is to ignore it and just look at the other 'Mark' values you posted above.

For CPU result, both the VD and the Laptop have CPU results about the same as a 6 year old desktop CPU. This class of desktop machine typically sells for $50 in Ebay now. So not exactly a spectacular result.

The disk results are interesting. Laptop again matches a 6 year old desktop (i.e. pretty poor), but I suspect the VD is doing some caching, and gets higher results. i.e. you might be partially measuring the cache speed, not the disk speed. So this might be worth some additional investigation. Or maybe the VD really does have a better disk sub-system.