Overclock.net › Forums › General Hardware › General Processor Discussions › What is the deal with AMD and Intel in a Linux environment?
New Posts  All Forums:Forum Nav:

What is the deal with AMD and Intel in a Linux environment? - Page 3

post #21 of 25
Quote:
Originally Posted by j3st3r View Post

This is such a useless post, dude. Is it moot because there is no difference in how Xeon and Opterons work inside of linux? Which is why I originally posted the benchmarks showing there CAN be a huge difference.

I don't think you fully understood his post, which was rather succinct. Basically, whichever CPU fits your needs is determined by what your workload is. There is no clearly defined 'best CPU' for all or even most workloads.
post #22 of 25
Quote:
Originally Posted by j3st3r View Post

This is such a useless post, dude. Is it moot because there is no difference in how Xeon and Opterons work inside of linux? Which is why I originally posted the benchmarks showing there CAN be a huge difference.
First, the post explained why you are running a parallel problem.... something that you must recognize now.

Also, you must understand your workload before anyone else can assist you with hardware recommendations.

As I have previously and repeatedly stated, most benchmarks nearly useless for HPC or custom software. Your software is not video encoding, a video game, Monte Carlo simulation, analytical biochemistry, financial quantitative analysis, hashing, equation solving, etc. (Common benchmarks basis.)

You are assuming the difference in the benchmarks are solely due to Xeon and Opteron in Linux. The source of the difference can be microarchitecture, interconnects, memory access, OS, compiler, and most importantly.... the benchmark/software itself. The differences in the benchmarks you posted may have absolutely no impact to you.... because your software is difference.

It's like asking "Which is better to drive in Mexico... a Ford F-350 or Porsche 911?" The answer depends on road conditions, route, expected weather, number of people, amount of luggage, etc. The what and where alone do not provide enough information.

With servers and HPC, you must really understand the requirements and workload before making decisions on hardware. If you do not, then you really are just guessing.
Edited by DuckieHo - 5/20/13 at 1:02pm
Once again...
(13 items)
 
  
CPUMotherboardGraphicsRAM
i7 920 [4.28GHz, HT] Asus P6T + Broadcom NetXtreme II VisionTek HD5850 [900/1200] + Galaxy GT240 2x4GB G.Skill Ripjaw X [1632 MHz] 
Hard DriveOSMonitorKeyboard
Intel X25-M 160GB + 3xRAID0 500GB 7200.12 Window 7 Pro 64 Acer H243H + Samsung 226BW XARMOR-U9BL  
PowerCaseMouseMouse Pad
Antec Truepower New 750W Li Lian PC-V2100 [10x120mm fans] Logitech G9 X-Trac Pro 
  hide details  
Reply
Once again...
(13 items)
 
  
CPUMotherboardGraphicsRAM
i7 920 [4.28GHz, HT] Asus P6T + Broadcom NetXtreme II VisionTek HD5850 [900/1200] + Galaxy GT240 2x4GB G.Skill Ripjaw X [1632 MHz] 
Hard DriveOSMonitorKeyboard
Intel X25-M 160GB + 3xRAID0 500GB 7200.12 Window 7 Pro 64 Acer H243H + Samsung 226BW XARMOR-U9BL  
PowerCaseMouseMouse Pad
Antec Truepower New 750W Li Lian PC-V2100 [10x120mm fans] Logitech G9 X-Trac Pro 
  hide details  
Reply
post #23 of 25
Thread Starter 
Your answers to my post have been nothing but kindergarten level analysis of the situation. You say the most basic things with no apparent background experience in operating software in both Linux and Windows environments. The software benchmark I posted, which I assume at this point you have not looked at, shows the same program being run in both Windows and Linux environments with a huge discrepancy in performance. If you could refrain come continually derailing this thread with your obvious software makes a difference banter that would be appreciated.

As I clearly said in the first post, I am looking for specific answers and benchmarks with software. Not troll baiting or fanboy answers. If you don't have experience, which clearly at this point I can confidently say you don't, then just don't post. Obviously software makes a difference. I am looking into reasons why the performance in that benchmark is so wide and what this means for the use of other software. Your analysis is basic at best and does nothing to answer specific questions. Basically you're saying this: "Given certain platforms, software will operate differently."

Well done. We are trying to take this 5 steps ahead of this train of thought.
post #24 of 25
Quote:
Originally Posted by j3st3r View Post

Your answers to my post have been nothing but kindergarten level analysis of the situation. You say the most basic things with no apparent background experience in operating software in both Linux and Windows environments. The software benchmark I posted, which I assume at this point you have not looked at, shows the same program being run in both Windows and Linux environments with a huge discrepancy in performance. If you could refrain come continually derailing this thread with your obvious software makes a difference banter that would be appreciated.

As I clearly said in the first post, I am looking for specific answers and benchmarks with software. Not troll baiting or fanboy answers. If you don't have experience, which clearly at this point I can confidently say you don't, then just don't post. Obviously software makes a difference. I am looking into reasons why the performance in that benchmark is so wide and what this means for the use of other software. Your analysis is basic at best and does nothing to answer specific questions. Basically you're saying this: "Given certain platforms, software will operate differently."

Well done. We are trying to take this 5 steps ahead of this train of thought.

Ahhh, I do have server experience... I did the analysis to place an ordered for $300K of servers just last year. I also have some HPC experience as well plus a background in computer engineering so I know quite a bit about OS design. I do performance profiling of software and do software/SQL optimization as part of my job.

If you want to point fingers... you didn't even realize that the software that you are support is master-slave parallel processing nor did you recognize the fundamental flaw in the design. (It's not time-sensitive so the storage subsystem should be good enough to cache everything for eventually processing. This will scale better and ALWAYS be more robust than attempting to have enough CPU power.) Have you done any analysis of the software you are trying to support? Again, I can tell you that no one will be able to provide you with the answer you are looking for which should be "Should I go with Opteron or Xeons for my requirements?"

You cannot advance if you didn't do the basic work.... how many performance systems have you designed before?

As for the NAS Parallel Benchmarks, why don't you just read what each benchmark stresses? If you care about these benchmarks (and you shouldn't), just delve into the source code, lib, and compiler....
http://www.nas.nasa.gov/publications/npb.html
Edited by DuckieHo - 5/20/13 at 9:27pm
Once again...
(13 items)
 
  
CPUMotherboardGraphicsRAM
i7 920 [4.28GHz, HT] Asus P6T + Broadcom NetXtreme II VisionTek HD5850 [900/1200] + Galaxy GT240 2x4GB G.Skill Ripjaw X [1632 MHz] 
Hard DriveOSMonitorKeyboard
Intel X25-M 160GB + 3xRAID0 500GB 7200.12 Window 7 Pro 64 Acer H243H + Samsung 226BW XARMOR-U9BL  
PowerCaseMouseMouse Pad
Antec Truepower New 750W Li Lian PC-V2100 [10x120mm fans] Logitech G9 X-Trac Pro 
  hide details  
Reply
Once again...
(13 items)
 
  
CPUMotherboardGraphicsRAM
i7 920 [4.28GHz, HT] Asus P6T + Broadcom NetXtreme II VisionTek HD5850 [900/1200] + Galaxy GT240 2x4GB G.Skill Ripjaw X [1632 MHz] 
Hard DriveOSMonitorKeyboard
Intel X25-M 160GB + 3xRAID0 500GB 7200.12 Window 7 Pro 64 Acer H243H + Samsung 226BW XARMOR-U9BL  
PowerCaseMouseMouse Pad
Antec Truepower New 750W Li Lian PC-V2100 [10x120mm fans] Logitech G9 X-Trac Pro 
  hide details  
Reply
post #25 of 25
Thread Starter 
It is not parallel. I have no clue where you're getting this assessment from. You're calling this program parallel in the same way you would call a switch with open flow parallel. It is simply NOT parallel. Especially considering so much load balancing is done by 2 pieces of separate hardware. You're using that term incorrectly and derailing this thread.

Second, the NAS benchmarks are the first page of the benchmarks entirety. Do yourself a favor and hit the 'next' button. Specifically, try to figure out why it is the x264 benchmarks show a huge gap in performance on Windows vs. Linux. As I clearly stated in the initial post, I am looking for specific data backed answers. Not your opinions on what I run.
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: General Processor Discussions
Overclock.net › Forums › General Hardware › General Processor Discussions › What is the deal with AMD and Intel in a Linux environment?