Overclock.net › Forums › Benchmarks › Overclock.net HWBot Team › HWBOT Prime
New Posts  All Forums:Forum Nav:

HWBOT Prime - Page 3

post #21 of 89
I think the current top score on HWBot Prime is Intel, and has done since day one. That score was crippled by low power CPUs and half threads because HWBotPrime doesn't let you set the number of threads.

Most people do not code for a specific architecture - only when you get down to the nitty gritty of optimisation do you consider the platform and use commands above SSE (such as AVX).

Take for example a benchmark based on pure floating point scenario, indicative of real world scientific computing:

Single thread:



Multi Thread:



Real world WinRAR Compression test:



Video Conversion:



Rendering:



Grid solver for scientific computing:



nBody simulation (coded up to SSE4):



Lets play a numbers game thumb.gif

HWBot Prime, speaking with the developer, is not coded for any specific platform. It's based off of Java for a start, just to make it platform/OS independent as much as possible - specifically for stuff like the Pi. He had no intention (as with many regular developers rather than high end developers) to code any particular way possible - just what made sense in the code and in the mathematics. If AMD or Intel want to design an arcitecture which fits the maths better, or fits the way the OS organises itself better, then that's up to them - that's not bias on the point of the developer.
Edited by borandi - 6/8/13 at 5:33am
post #22 of 89
Thread Starter 
Quote:
Originally Posted by borandi View Post

I think the current top score on HWBot Prime is Intel, and has done since day one. That score was crippled by low power CPUs and half threads because HWBotPrime doesn't let you set the number of threads.

Most people do not code for a specific architecture - only when you get down to the nitty gritty of optimisation do you consider the platform and use commands above SSE (such as AVX).

Take for example a benchmark based on pure floating point scenario, indicative of real world scientific computing:

Single thread:

Multi Thread:

Real world WinRAR Compression test:

Video Conversion:

Rendering:

Grid solver for scientific computing:

nBody simulation (coded up to SSE4):

Lets play a numbers game thumb.gif

HWBot Prime, speaking with the developer, is not coded for any specific platform. It's based off of Java for a start, just to make it platform/OS independent as much as possible - specifically for stuff like the Pi. He had no intention (as with many regular developers rather than high end developers) to code any particular way possible - just what made sense in the code and in the mathematics. If AMD or Intel want to design an arcitecture which fits the maths better, or fits the way the OS organises itself better, then that's up to them - that's not bias on the point of the developer.
Way to post a bunch of irrelevant crap to the point. Did you read my posts at all? I am going to say you didn't based on this post. First if you managed to spend more than 2sec looking at the HWBOT Prime rankings you'd see that yes the first spot is Intel but (just to help you this time ---> in bold) It has 32 cores. You know, when you start your post with ignorance it makes the rest laughable. We have all seen your benches before. You think posting them for the 899th time proves more than it did the first? Besides you didn't disprove my point, or question actually. Do you think those benches were programmed and tested on AMD hardware? So how can you say it is an accurate test of ACTUAL performance. To help with the point, I will use a car analogy: Take 2 cars both with equal horsepower, one a drag car the other a road-course car. Now take them both to Lemans and set them free. We all know the drag car is gonna get stomped. But does that mean its performance is sub-par? On a road course, Yes. But go to the drag strip and see what happens. Point is: How can we be sure of true performance of either manufacture? Of course real world application, as I stated in a previous post, does that and Intel wins hands down. But say the development of software changes. Do you think Intel will still fair as well?
post #23 of 89
Quote:
Originally Posted by Durquavian View Post

Quote:
Originally Posted by borandi View Post

I think the current top score on HWBot Prime is Intel, and has done since day one. That score was crippled by low power CPUs and half threads because HWBotPrime doesn't let you set the number of threads.

Most people do not code for a specific architecture - only when you get down to the nitty gritty of optimisation do you consider the platform and use commands above SSE (such as AVX).

Take for example a benchmark based on pure floating point scenario, indicative of real world scientific computing:

Single thread:

Multi Thread:

Real world WinRAR Compression test:

Video Conversion:

Rendering:

Grid solver for scientific computing:

nBody simulation (coded up to SSE4):

Lets play a numbers game thumb.gif

HWBot Prime, speaking with the developer, is not coded for any specific platform. It's based off of Java for a start, just to make it platform/OS independent as much as possible - specifically for stuff like the Pi. He had no intention (as with many regular developers rather than high end developers) to code any particular way possible - just what made sense in the code and in the mathematics. If AMD or Intel want to design an arcitecture which fits the maths better, or fits the way the OS organises itself better, then that's up to them - that's not bias on the point of the developer.
Way to post a bunch of irrelevant crap to the point. Did you read my posts at all? I am going to say you didn't based on this post. First if you managed to spend more than 2sec looking at the HWBOT Prime rankings you'd see that yes the first spot is Intel but (just to help you this time ---> in bold) It has 32 cores. You know, when you start your post with ignorance it makes the rest laughable. We have all seen your benches before. You think posting them for the 899th time proves more than it did the first? Besides you didn't disprove my point, or question actually. Do you think those benches were programmed and tested on AMD hardware? So how can you say it is an accurate test of ACTUAL performance. To help with the point, I will use a car analogy: Take 2 cars both with equal horsepower, one a drag car the other a road-course car. Now take them both to Lemans and set them free. We all know the drag car is gonna get stomped. But does that mean its performance is sub-par? On a road course, Yes. But go to the drag strip and see what happens. Point is: How can we be sure of true performance of either manufacture? Of course real world application, as I stated in a previous post, does that and Intel wins hands down. But say the development of software changes. Do you think Intel will still fair as well?

Let's correct you.

Firstly, if you spent 2 seconds looking at who posted that score, you'll know that it was me who set that score. I know how many cores it has, I also understand the limitations of the benchmarks - I've been at the top end of the Hwbot enthusiast league for years and now I'm in a top pro OC team.

When you make a claim that the benchmark favors Intel, and it wasn't tested on amd, guess what - some of the benchmark results I posted were from benchmarks I have written, written on amd systems. Care to argue with the benchmark writer about being Intel focused? All you are doing is interpreting benchmark results without any hard data or evidence and no respectable scientific analyst would accept your interpretations

Mod Edit: Removed quoted images for readability!
post #24 of 89
Thread Starter 
Quote:
Originally Posted by borandi View Post

Let's correct you.

Firstly, if you spent 2 seconds looking at who posted that score, you'll know that it was me who set that score. I know how many cores it has, I also understand the limitations of the benchmarks - I've been at the top end of the Hwbot enthusiast league for years and now I'm in a top pro OC team.

When you make a claim that the benchmark favors Intel, and it wasn't tested on amd, guess what - some of the benchmark results I posted were from benchmarks I have written, written on amd systems. Care to argue with the benchmark writer about being Intel focused? All you are doing is interpreting benchmark results without any hard data or evidence and no respectable scientific analyst would accept your interpretations
What hardware did you write the code... On a napkin, doubt it. Your short sighted responses don't help either. So you knew you were first on that ranking and yet left out the very important detail of 32 cores. If you take a min or two to pull your head out of the Intel sand it is buried you would see my point. But alas another fanboy with no rational response to the topic on hand.
post #25 of 89
Quote:
Originally Posted by borandi View Post

I think the current top score on HWBot Prime is Intel, and has done since day one. That score was crippled by low power CPUs and half threads because HWBotPrime doesn't let you set the number of threads.

Most people do not code for a specific architecture - only when you get down to the nitty gritty of optimisation do you consider the platform and use commands above SSE (such as AVX).

Take for example a benchmark based on pure floating point scenario, indicative of real world scientific computing:

Single thread:

Multi Thread:

Real world WinRAR Compression test:

Video Conversion:

Rendering:

Grid solver for scientific computing:

nBody simulation (coded up to SSE4):

Lets play a numbers game thumb.gif

HWBot Prime, speaking with the developer, is not coded for any specific platform. It's based off of Java for a start, just to make it platform/OS independent as much as possible - specifically for stuff like the Pi. He had no intention (as with many regular developers rather than high end developers) to code any particular way possible - just what made sense in the code and in the mathematics. If AMD or Intel want to design an arcitecture which fits the maths better, or fits the way the OS organises itself better, then that's up to them - that's not bias on the point of the developer.

I must agree. General compilation strategy is generic. It is good practice to support optimizations where available and available CPU flags can be detected and utilized at run time. My strategy is of course different if I'm compiling specifically for a single machine, such as my own. Then I optimize for that given processor as much as possible, given the requirements of the program.

One thing that occurred to me about HWBot Prime, is that while being Java based makes it platform/OS independent, it also adds more variables in to the equation. For instance, I have OpenJDK on my system and without trying Oracle's JRE I wouldn't know if it will produce better results with this benchmark. Even a different release of either might be better or worse. If I were a cheater, I could compile a custom version of OpenJDK and perhaps gain some performance with that. Those minor points of concern aside, I'm grateful to the developer of HWBot Prime for making an OS independent benchmark. I'm often left out of the benchmark game, because I refuse to run Windows on anything other than a VM wink.gif Of course, it is my stubborn nature that limits benchmarks available to me rolleyes.gif
deathstar
(26 items)
 
  
CPUMotherboardGraphicsRAM
FX-8350 @ 4.5 GHz Asus M5A99X EVO R2.0 EVGA GeForce GTX 670 (02G-P4-2670-BR) Kingston HyperX Blu KHX1600C10D3B1K2/16G X 2 
Hard DriveHard DriveHard DriveOptical Drive
Samsung 840 Pro WD WD2500HHTZ VelociRaptor WDC DC10EADS-00M LG CD/DVD Writer 
CoolingCoolingCoolingCooling
Cooler Master 120XL Cooler Master MegaFlow 200mm Cooler Master MegaFlow 200mm Cooler Master Blade Master 120mm 
CoolingCoolingCoolingOS
Cooler Master Blade Master 120mm COUGAR CF-V12HP Vortex  COUGAR CF-V12HP Vortex  Ubuntu Studio 13.04 
MonitorMonitorKeyboardPower
ViewSonic VX2450-wm ViewSonic VX2450-wm WASD V1 w/Cherry Blues Cooler Master Silent Pro M 850W 
CaseMouseAudioOther
Cooler Master HAF 912 Microsoft L2 Sculpt Touch Mouse 6PL-00003 Creative Inspire P5800 - 5.1 Speakers Inland 05322 Dual LCD Desk Mount 
OtherOther
Targus ACB10US Bluetooth Adapter APC Back-UPS XS-1000 
  hide details  
Reply
deathstar
(26 items)
 
  
CPUMotherboardGraphicsRAM
FX-8350 @ 4.5 GHz Asus M5A99X EVO R2.0 EVGA GeForce GTX 670 (02G-P4-2670-BR) Kingston HyperX Blu KHX1600C10D3B1K2/16G X 2 
Hard DriveHard DriveHard DriveOptical Drive
Samsung 840 Pro WD WD2500HHTZ VelociRaptor WDC DC10EADS-00M LG CD/DVD Writer 
CoolingCoolingCoolingCooling
Cooler Master 120XL Cooler Master MegaFlow 200mm Cooler Master MegaFlow 200mm Cooler Master Blade Master 120mm 
CoolingCoolingCoolingOS
Cooler Master Blade Master 120mm COUGAR CF-V12HP Vortex  COUGAR CF-V12HP Vortex  Ubuntu Studio 13.04 
MonitorMonitorKeyboardPower
ViewSonic VX2450-wm ViewSonic VX2450-wm WASD V1 w/Cherry Blues Cooler Master Silent Pro M 850W 
CaseMouseAudioOther
Cooler Master HAF 912 Microsoft L2 Sculpt Touch Mouse 6PL-00003 Creative Inspire P5800 - 5.1 Speakers Inland 05322 Dual LCD Desk Mount 
OtherOther
Targus ACB10US Bluetooth Adapter APC Back-UPS XS-1000 
  hide details  
Reply
post #26 of 89
Just out of curiosity I've popped out a calculator to check my efficiency:

mega__Man - FX8350 @ 5528 - 8C/8T > 6966.53 -> Eff 0.1575
GENiEBEN - L5518 @ 3712 - 4C/8T > 4794.01 -> Eff. 0.1614 (which translates to ~7100 at the same clocks of the BD).

Waiting for 5.5 Gulfy's or 6G+ Ivy/Nancy before I can call this an AMD benchie smile.gif
Quote:
Originally Posted by FunkyPresident View Post

I could compile a custom version of OpenJDK and perhaps gain some performance with that.

Nothing illegal btw, you could just do that. If you were a cheater you'd have an easy task. as both JAVA/NET suffer from no code obfuscation.
Edited by GENiEBEN - 6/8/13 at 10:36am
post #27 of 89
Quote:
Originally Posted by GENiEBEN View Post

Just out of curiosity I've popped out a calculator to check my efficiency:

mega__Man - FX8350 @ 5528 - 8C/8T > 6966.53 -> Eff 0.1575
GENiEBEN - L5518 @ 3712 - 4C/8T > 4794.01 -> Eff. 0.1614 (which translates to ~7100 at the same clocks of the BD).

Waiting for 5.5 Gulfy's or 6G+ Ivy/Nancy before I can call this an AMD benchie smile.gif

There are 4370Mhz 8350s (Close to stock) scoring higher than 5178Mhz 3770ks.

6Ghz on Ivy requires Phase or LN2, and it'll only get them back in the top 20, not give them the lead. Even the 5.5Ghz 8350 is on "only" water (they are obviously not 24/7 stable at that speed).
Forge
(17 items)
 
Forge-LT
(7 items)
 
 
CPUMotherboardGraphicsGraphics
Intel i7-5960X (4.625Ghz) ASUS X99-DELUXE/U3.1 EVGA 1080ti SC2 Hybrid EVGA 1080ti SC2 Hybrid 
RAMHard DriveCoolingOS
64GB Corsair Dominator Platinum (3000Mhz 8x8GB) Samsung 950 Pro NVMe 512GB EK Predator 240 Windows 10 Enterprise x64 
MonitorKeyboardPowerCase
2x Acer XR341CK Corsair Vengeance K70 RGB Corsair AX1200 Corsair Graphite 780T 
MouseAudioAudioAudio
Corsair Vengeance M65 RGB Sennheiser HD700 Sound Blaster AE-5 Audio Technica AT4040 
Audio
30ART Mic Tube Amp 
CPUMotherboardGraphicsRAM
i7-4720HQ UX501JW-UB71T GTX 960m 16GB 1600 9-9-9-27 
Hard DriveOSMonitor
512GB PCI-e SSD Windows 10 Pro 4k IPS 
  hide details  
Reply
Forge
(17 items)
 
Forge-LT
(7 items)
 
 
CPUMotherboardGraphicsGraphics
Intel i7-5960X (4.625Ghz) ASUS X99-DELUXE/U3.1 EVGA 1080ti SC2 Hybrid EVGA 1080ti SC2 Hybrid 
RAMHard DriveCoolingOS
64GB Corsair Dominator Platinum (3000Mhz 8x8GB) Samsung 950 Pro NVMe 512GB EK Predator 240 Windows 10 Enterprise x64 
MonitorKeyboardPowerCase
2x Acer XR341CK Corsair Vengeance K70 RGB Corsair AX1200 Corsair Graphite 780T 
MouseAudioAudioAudio
Corsair Vengeance M65 RGB Sennheiser HD700 Sound Blaster AE-5 Audio Technica AT4040 
Audio
30ART Mic Tube Amp 
CPUMotherboardGraphicsRAM
i7-4720HQ UX501JW-UB71T GTX 960m 16GB 1600 9-9-9-27 
Hard DriveOSMonitor
512GB PCI-e SSD Windows 10 Pro 4k IPS 
  hide details  
Reply
post #28 of 89
Quote:
Originally Posted by GENiEBEN View Post

Nothing illegal btw, you could just do that. If you were a cheater you'd have an easy task. as both JAVA/NET suffer from no code obfuscation.

Compiling an optimized OpenJDK could be interesting. Not for this benchmark, but for say a Glassfish server. I've taken a liking to Glassfish and JavaEE style web applications. If a custom compiled JDK would improve performance on a given platform, that would be excellent. I've been setting up my Glassfish servers as VMs, but I've been meaning to set one up on this machine to play with JCuda anyway.

Also in the interest in civility, allow me to apologize for pegging you as VS user and other rudeness. I had to remind myself, that these discussions should remain civil.

It seems like more and more I have to write for MS based systems, to the point that I keep Windows VMs on my machines. I may set up MinGW on one of my VMs, since I'm used to GCC and I am obviously not a fan of MS products. GCC has always been a given on my systems. Of course, it is possible to cross-compile Windows applications from Linux, but it is a bit of a hassle. However, as I've learned the hard way, writing for Windows is were the money is at. No matter how much I despise it as an OS. SMH

As far as all this Intel vs AMD stuff. I have to say, meh. It is of course in everyone's interest for them to be competitive with each other. I bought this 8350, because it was hard for me to pass up a 8 core CPU for under $200. Plus, I've always like underdogs. I'm pretty happy with it. Really to me, whatever get's the job done is cool. I hope both companies step up their efforts in the coming years.

HWBot Prime seems to be a pretty fair benchmark and is looking good for AMD. Not bad at all considering the price of the Intel chips they're competing against, power consumption aside. Let's be honest, OCers aren't exactly worried about being "green", when it comes to power consumption anyway.
deathstar
(26 items)
 
  
CPUMotherboardGraphicsRAM
FX-8350 @ 4.5 GHz Asus M5A99X EVO R2.0 EVGA GeForce GTX 670 (02G-P4-2670-BR) Kingston HyperX Blu KHX1600C10D3B1K2/16G X 2 
Hard DriveHard DriveHard DriveOptical Drive
Samsung 840 Pro WD WD2500HHTZ VelociRaptor WDC DC10EADS-00M LG CD/DVD Writer 
CoolingCoolingCoolingCooling
Cooler Master 120XL Cooler Master MegaFlow 200mm Cooler Master MegaFlow 200mm Cooler Master Blade Master 120mm 
CoolingCoolingCoolingOS
Cooler Master Blade Master 120mm COUGAR CF-V12HP Vortex  COUGAR CF-V12HP Vortex  Ubuntu Studio 13.04 
MonitorMonitorKeyboardPower
ViewSonic VX2450-wm ViewSonic VX2450-wm WASD V1 w/Cherry Blues Cooler Master Silent Pro M 850W 
CaseMouseAudioOther
Cooler Master HAF 912 Microsoft L2 Sculpt Touch Mouse 6PL-00003 Creative Inspire P5800 - 5.1 Speakers Inland 05322 Dual LCD Desk Mount 
OtherOther
Targus ACB10US Bluetooth Adapter APC Back-UPS XS-1000 
  hide details  
Reply
deathstar
(26 items)
 
  
CPUMotherboardGraphicsRAM
FX-8350 @ 4.5 GHz Asus M5A99X EVO R2.0 EVGA GeForce GTX 670 (02G-P4-2670-BR) Kingston HyperX Blu KHX1600C10D3B1K2/16G X 2 
Hard DriveHard DriveHard DriveOptical Drive
Samsung 840 Pro WD WD2500HHTZ VelociRaptor WDC DC10EADS-00M LG CD/DVD Writer 
CoolingCoolingCoolingCooling
Cooler Master 120XL Cooler Master MegaFlow 200mm Cooler Master MegaFlow 200mm Cooler Master Blade Master 120mm 
CoolingCoolingCoolingOS
Cooler Master Blade Master 120mm COUGAR CF-V12HP Vortex  COUGAR CF-V12HP Vortex  Ubuntu Studio 13.04 
MonitorMonitorKeyboardPower
ViewSonic VX2450-wm ViewSonic VX2450-wm WASD V1 w/Cherry Blues Cooler Master Silent Pro M 850W 
CaseMouseAudioOther
Cooler Master HAF 912 Microsoft L2 Sculpt Touch Mouse 6PL-00003 Creative Inspire P5800 - 5.1 Speakers Inland 05322 Dual LCD Desk Mount 
OtherOther
Targus ACB10US Bluetooth Adapter APC Back-UPS XS-1000 
  hide details  
Reply
post #29 of 89
Quote:
Originally Posted by Durquavian View Post

What hardware did you write the code... On a napkin, doubt it.

Excuse me? Urinating against the wind now? Or just a series of non-sequiturs?
Quote:
So you knew you were first on that ranking and yet left out the very important detail of 32 cores.

So you know you can actually read, right? Especially when I say
Quote:
Originally Posted by borandi View Post

That score was crippled by low power CPUs and half threads because HWBotPrime doesn't let you set the number of threads.

Also,
Quote:
If you take a min or two to pull your head out of the Intel sand it is buried you would see my point. But alas another fanboy with no rational response to the topic on hand.

If you really want to get your point across, then trying to crap on everyone that actually works in this field is doing a great job for you. I applaud your efforts. If you actually spend five minutes to find out who is responding to your comments, then statements like the one above would be laughable - *at best*.

But this is the internet , and people who want to have an opinion regardless of interpretation skills and quality are always the ones in the right, regardless of numbers, "facts" or evidence. I bow to your superior intellect on the matter.
post #30 of 89
Quote:
Originally Posted by FunkyPresident View Post

Also in the interest in civility, allow me to apologize for pegging you as VS user and other rudeness. I had to remind myself, that these discussions should remain civil.

No offense taken, I use the proper tools for the given task, if that involves VS so be it.
Quote:
Originally Posted by FunkyPresident View Post

It seems like more and more I have to write for MS based systems, to the point that I keep Windows VMs on my machines. I may set up MinGW on one of my VMs, since I'm used to GCC and I am obviously not a fan of MS products. GCC has always been a given on my systems. Of course, it is possible to cross-compile Windows applications from Linux, but it is a bit of a hassle. However, as I've learned the hard way, writing for Windows is were the money is at. No matter how much I despise it as an OS. SMH.

Well you can bet the money's to be made from Win applications smile.gif
Quote:
Originally Posted by FunkyPresident View Post

As far as all this Intel vs AMD stuff. I have to say, meh. It is of course in everyone's interest for them to be competitive with each other. I bought this 8350, because it was hard for me to pass up a 8 core CPU for under $200. Plus, I've always like underdogs. I'm pretty happy with it. Really to me, whatever get's the job done is cool. I hope both companies step up their efforts in the coming years..

Couldn't care about who's winning either, as someone who benchmarked in excess of 200 different cpus I can only say: bring me more of both biggrin.gif
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Overclock.net HWBot Team
Overclock.net › Forums › Benchmarks › Overclock.net HWBot Team › HWBOT Prime