netkas.org forum
June 23, 2017, 01:11:03 AM *
Welcome, Guest. Please login or register.
Did you miss your activation email?

Login with username, password and session length
News: Information for registering users http://forum.netkas.org/index.php/topic,2246.0.html
 
   Home   Help Search Login Register  
Pages: [1] 2
  Print  
Author Topic: 5K testing with Pascal  (Read 1406 times)
Rominator
Hero Member
*****
Offline Offline

Posts: 2123



« on: April 13, 2017, 12:02:10 AM »

So for some reason everyone at MR keeps testing at *Default* settings with Valley.

Sort of like testing a Ferrari on 30 MPH city streets.

SO I took the cards out on an Autobahn

5K in Valley at Ultra

The 1080 does 50% better than Titan Maxwell, the Titan Pascal does 100% better.

http://www.macvidcards.com/blog
« Last Edit: April 13, 2017, 12:13:41 AM by Rominator » Logged

Before asking a question, check your "Personal Settings" and be sure that you have "Brain Services" set to "On".
Asgorath
Sr. Member
****
Offline Offline

Posts: 252


« Reply #1 on: April 13, 2017, 01:16:00 AM »

Yeah, that's more like it.  As expected it took people less than a day to post their silly 1600x900 results and complain that the drivers were shit.
Logged
Rominator
Hero Member
*****
Offline Offline

Posts: 2123



« Reply #2 on: April 13, 2017, 01:54:36 AM »

If you look, the difference between the two Titans is quite amazing.

Not just final score but ALL of the numbers are pretty much doubled.

With a load not centered on CPU, these new cards and drivers FLY!

Congrats to Nvidia and the Mac Driver team.

And to Apple, for hopefully pulling their heads out of...
« Last Edit: April 13, 2017, 03:18:36 AM by Rominator » Logged

Before asking a question, check your "Personal Settings" and be sure that you have "Brain Services" set to "On".
karaluchy
Newbie

Offline Offline

Posts: 40


« Reply #3 on: April 13, 2017, 07:00:30 AM »

Yeah, that's more like it.  As expected it took people less than a day to post their silly 1600x900 results and complain that the drivers were shit.

No need to be impolite, I'm one of them  Wink
We asked, we doubted and you guys with greater knowledge responeded and talked us through.
Thanks again for the benchmarks and your time... ;P
Logged
lastExile
Global Moderator
Newbie

Offline Offline

Posts: 36



« Reply #4 on: April 13, 2017, 09:18:12 AM »

Yeah, that's more like it.  As expected it took people less than a day to post their silly 1600x900 results and complain that the drivers were shit.

is that the first gen titan x pascal or the new one ?
Logged

Report all spam and i will take care of it
Spacedust
Full Member
***
Offline Offline

Posts: 248


« Reply #5 on: April 13, 2017, 10:10:21 AM »

So the CPU is still not limiting the performance. These Xeons will last forerver. You would have much less noise if you use X5675's instead of these overheating X5690's.
Logged
Spacedust
Full Member
***
Offline Offline

Posts: 248


« Reply #6 on: April 13, 2017, 10:29:44 AM »

New article on Barefeats: http://barefeats.com/cmp_pascal.html

Dual 1080's works very well!
Logged
Asgorath
Sr. Member
****
Offline Offline

Posts: 252


« Reply #7 on: April 13, 2017, 02:39:10 PM »

So the CPU is still not limiting the performance. These Xeons will last forerver. You would have much less noise if you use X5675's instead of these overheating X5690's.

Do you have a 5K monitor and are comfortable playing games at 10-20 FPS?  If not, then the CPUs are still a limiting factor for games.  The CPU technology in the cMP is coming up on 9 years old.  I was shocked at how much faster my Haswell-based Hackintosh was compared with my old 2010 Mac Pro, and I haven't looked back.
Logged
DarthnVader
Newbie

Offline Offline

Posts: 12


« Reply #8 on: April 13, 2017, 03:23:36 PM »

So for some reason everyone at MR keeps testing at *Default* settings with Valley.

Sort of like testing a Ferrari on 30 MPH city streets.

SO I took the cards out on an Autobahn

5K in Valley at Ultra

The 1080 does 50% better than Titan Maxwell, the Titan Pascal does 100% better.

http://www.macvidcards.com/blog


Great work, looks like the cMP may still have some legs yet. If I read your benchmarks right, you're using a 12 core 24 thread cMP, I'm curious how many cores/threads Valley can utilize?

Most modern games can take advantage of 6 cores, I think.

 
Logged
Spacedust
Full Member
***
Offline Offline

Posts: 248


« Reply #9 on: April 13, 2017, 03:43:17 PM »

Do you have a 5K monitor and are comfortable playing games at 10-20 FPS?  If not, then the CPUs are still a limiting factor for games.  The CPU technology in the cMP is coming up on 9 years old.  I was shocked at how much faster my Haswell-based Hackintosh was compared with my old 2010 Mac Pro, and I haven't looked back.

No, but I got a second 4K 24" display and simply cannot use it for normal work.

Everything is ultra small (especially under Windows 7), so I'm using Apple Cinema Display 23" and it's 1920x1200 resolution.

Such displays are only good for gaming, video editing and nothing more.
Logged
Fl0r!an
Sr. Member
****
Offline Offline

Posts: 346


« Reply #10 on: April 13, 2017, 04:38:27 PM »

Obviously it depends on every application whether it's gonna be CPU limited in the cMP. Benchmarking at 1600x900 is stupid, no one buys a GTX 1080/Titan X to game at such a resolution.

On the other hand there's no GPU yet which can handle 5K gaming. It's still good to see that the GPUs behave as expected when they're pushed to their limits.

Many OS X games are heavily CPU bound, caused by high CPU overhead in drivers and/or bad ports, which was very obvious when I compared my old cMP3,1 vs. a Nehalem and a Skylake Hackintosh. Even a slow GPU like my old R9 280 can get CPU limited in my Nehalem rig at 1080p or 1440p, not talking about monsters like a 1080 or Titan Xp.

This shouldn't apply for most professional GPGPU applications though.

@Spacedust: What about HiDPI mode? Even Windows 7 is able to apply nice vector scaling...
Logged
Rominator
Hero Member
*****
Offline Offline

Posts: 2123



« Reply #11 on: April 13, 2017, 08:15:58 PM »

To be clear, I only went full 5K to show these results

99% of time I use that Dell at 2560x1600 via HiDPI scaling, it looks incredible

Logged

Before asking a question, check your "Personal Settings" and be sure that you have "Brain Services" set to "On".
Spacedust
Full Member
***
Offline Offline

Posts: 248


« Reply #12 on: April 13, 2017, 09:56:35 PM »

@Spacedust: What about HiDPI mode? Even Windows 7 is able to apply nice vector scaling...

On Windows 7 you can only enlarge fonts up to 150% and it doesn't work well for older software:

Logged
Squishy Tia
Full Member
***
Offline Offline

Posts: 100



« Reply #13 on: April 14, 2017, 03:15:20 AM »

So for some reason everyone at MR keeps testing at *Default* settings with Valley.

Sort of like testing a Ferrari on 30 MPH city streets.

SO I took the cards out on an Autobahn

5K in Valley at Ultra

The 1080 does 50% better than Titan Maxwell, the Titan Pascal does 100% better.

http://www.macvidcards.com/blog


This means that the 1080 Ti should come close to or matching those numbers for some games, since it's really just a Titan minus a few CUDA cores, 1 GB VRAM, but with a wider bus.
Logged
Rominator
Hero Member
*****
Offline Offline

Posts: 2123



« Reply #14 on: April 14, 2017, 07:38:11 AM »

Got the 1080Ti

More tomorrow
Logged

Before asking a question, check your "Personal Settings" and be sure that you have "Brain Services" set to "On".
Pages: [1] 2
  Print  
 
Jump to:  

Powered by MySQL Powered by PHP Powered by SMF 1.1.21 | SMF © 2015, Simple Machines
SMFAds for Free Forums
Valid XHTML 1.0! Valid CSS!