Topic: JRMark -- JRiver's Benchmark Tool (Read 239398 times)

JimH · « **on:** October 14, 2009, 05:13:40 pm »

Matt added a Benchmark program in build 83 that measures overall system performance. It gives you a way to compare machines. It also gives us a way to measure changes in the code and look for new ways to enhance performance.

It's under MC's Help menu.

Your results will change over time for several reasons:

1. A faster machine -- overclocking, for example
2. A newer version of MC with performance improvements
3. Power settings of the machine
4. Other programs running

STANDINGS:
~~Best: glynor JRMark (version 14.0.83): 2867~~
Best: Matt on MC16: 4376

Worst: rick.ca JRMark (version 14.0.83): 264

Quote from: Matt on October 14, 2009, 05:00:19 pm

10. NEW: Added 'Benchmark' tool to measure system performance and help diagnose performance problems. (Help > Benchmark...)

My office Core2 Quad at 2.66 GHz

Running 'Math' benchmark...
   Single-threaded integer math... 4.467 seconds
   Single-threaded floating point math... 2.663 seconds
   Multi-threaded integer math... 2.288 seconds
   Multi-threaded mixed math... 1.407 seconds
Score: 1755

Running 'Image' benchmark...
   Image creation / destruction... 1.626 seconds
   Flood filling... 1.659 seconds
   Direct copying... 2.076 seconds
   Small renders... 3.203 seconds
   Bilinear rendering... 1.555 seconds
   Bicubic rendering... 1.562 seconds
Score: 1883

Running 'Database' benchmark...
   Create database... 0.929 seconds
   Populate database... 4.197 seconds
   Save database... 0.212 seconds
   Reload database... 0.154 seconds
   Search database... 3.494 seconds
   Sort database... 3.311 seconds
   Group database... 2.931 seconds
Score: 1412

JRMark (version 14.0.83): 1684

gappie · « **Reply #1 on:** October 14, 2009, 05:16:03 pm »

Quote from: JimH on October 14, 2009, 05:13:40 pm

JRMark (version 14.0.83): 1684

JRMark (version 14.0.83): 937

hope its not to bad

gab

ps:

Quote

2. Fixed: Theater View themes wouldn't show root items (only applies to build 82).

works nicely again, just deleted my aa folder.

JimH · « **Reply #2 on:** October 14, 2009, 05:18:20 pm »

I think yours is better than mine.

gappie · « **Reply #3 on:** October 14, 2009, 05:22:36 pm »

Quote from: JimH on October 14, 2009, 05:18:20 pm

I think yours is better than mine.

i dont think so.. though i have no idea what i talk about..
here is the rest, all times are longer on my machine..

=== Running Benchmarks (please do not interrupt) ===

Running 'Math' benchmark...
Single-threaded integer math... 6.059 seconds
Single-threaded floating point math... 3.796 seconds
Multi-threaded integer math... 6.088 seconds
Multi-threaded mixed math... 3.788 seconds
Score: 963

Running 'Image' benchmark...
Image creation / destruction... 2.806 seconds
Flood filling... 3.182 seconds
Direct copying... 2.722 seconds
Small renders... 4.897 seconds
Bilinear rendering... 4.659 seconds
Bicubic rendering... 4.640 seconds
Score: 960

Running 'Database' benchmark...
Create database... 2.002 seconds
Populate database... 5.725 seconds
Save database... 0.442 seconds
Reload database... 0.275 seconds
Search database... 4.929 seconds
Sort database... 6.041 seconds
Group database... 5.130 seconds
Score: 876

JRMark (version 14.0.83): 933

Matt · « **Reply #4 on:** October 14, 2009, 05:24:17 pm »

Higher numbers are better.

Here's my work machine (Core Duo @ 2.88):

=== Running Benchmarks (please do not interrupt) ===

Running 'Math' benchmark...
   Single-threaded integer math... 3.953 seconds
   Single-threaded floating point math... 2.485 seconds
   Multi-threaded integer math... 3.908 seconds
   Multi-threaded mixed math... 2.451 seconds
Score: 1485

Running 'Image' benchmark...
   Image creation / destruction... 1.566 seconds
   Flood filling... 2.155 seconds
   Direct copying... 2.158 seconds
   Small renders... 3.306 seconds
   Bilinear rendering... 3.032 seconds
   Bicubic rendering... 3.009 seconds
Score: 1445

Running 'Database' benchmark...
   Create database... 1.043 seconds
   Populate database... 3.519 seconds
   Save database... 0.182 seconds
   Reload database... 0.090 seconds
   Search database... 3.396 seconds
   Sort database... 3.257 seconds
   Group database... 3.022 seconds
Score: 1482

JRMark (version 14.0.83): 1470

Matt · « **Reply #5 on:** October 14, 2009, 05:25:28 pm »

Jim proposed calling the benchmark TwainMarks.

JimH · « **Reply #6 on:** October 14, 2009, 05:27:52 pm »

I lost.

gappie · « **Reply #7 on:** October 14, 2009, 05:38:44 pm »

TwainMarks is funny, the question is who will be the twain at the and of this thread.

JimH · « **Reply #8 on:** October 14, 2009, 05:42:38 pm »

Quote from: gappie on October 14, 2009, 05:38:44 pm

.. the question is who will be the twain at the and of this thread.

Hmmm?

Twain means two. I think "Mark Twain" meant two fathoms when the steamboat was sounding the depth as they crossed shallow places.

Matt · « **Reply #9 on:** October 14, 2009, 05:47:58 pm »

Here's an i7 @ 3.2

=== Running Benchmarks (please do not interrupt) ===

Running 'Math' benchmark...
   Single-threaded integer math... 4.161 seconds
   Single-threaded floating point math... 2.345 seconds
   Multi-threaded integer math..1.210 seconds
   Multi-threaded mixed math... 0.825 seconds
Score: 2225

Running 'Image' benchmark...
   Image creation / destruction... 0.907 seconds
   Flood filling... 0.450 seconds
   Direct copying... 0.543 seconds
   Small renders... 2.595 seconds
   Bilinear rendering... 1.247 seconds
   Bicubic rendering... 1.207 seconds
Score: 3166

Running 'Database' benchmark...
   Create database... 0.708 seconds
   Populate database... 2.495 seconds
   Save database... 0.131 seconds
   Reload database... 0.058 seconds
   Search database... 2.436 seconds
   Sort database... 2.637 seconds
   Group database... 2.269 seconds
Score: 2003

JRMark (version 14.0.83): 2465

glynor · « **Reply #10 on:** October 14, 2009, 05:53:18 pm »

Here's my HTPC (Core 2 Quad Q9550 @ 3.5GHz, 4GB RAM, AMD HD 4870 GPU):

=== Running Benchmarks (please do not interrupt) ===

Running 'Math' benchmark...
Single-threaded integer math... 3.945 seconds
Single-threaded floating point math... 2.308 seconds
Multi-threaded integer math... 2.036 seconds
Multi-threaded mixed math... 1.271 seconds
Score: 1988

Running 'Image' benchmark...
Image creation / destruction... 1.210 seconds
Flood filling... 1.472 seconds
Direct copying... 1.920 seconds
Small renders... 2.602 seconds
Bilinear rendering... 1.408 seconds
Bicubic rendering... 1.418 seconds
Score: 2193

Running 'Database' benchmark...
Create database... 0.798 seconds
Populate database... 3.947 seconds
Save database... 0.341 seconds
Reload database... 0.082 seconds
Search database... 3.060 seconds
Sort database... 3.648 seconds
Group database... 2.616 seconds
Score: 1484

JRMark (version 14.0.83): 1888

glynor · « **Reply #11 on:** October 14, 2009, 05:57:10 pm »

And here's my server machine in the basement (Core i5 750 @ 4.0GHz, 4GB RAM, AMD HD 3850):

=== Running Benchmarks (please do not interrupt) ===

Running 'Math' benchmark...
Single-threaded integer math... 3.329 seconds
Single-threaded floating point math... 1.876 seconds
Multi-threaded integer math... 1.723 seconds
Multi-threaded mixed math... 1.008 seconds
Score: 2394

Running 'Image' benchmark...
Image creation / destruction... 0.700 seconds
Flood filling... 0.470 seconds
Direct copying... 0.676 seconds
Small renders... 2.105 seconds
Bilinear rendering... 0.985 seconds
Bicubic rendering... 0.975 seconds
Score: 3722

Running 'Database' benchmark...
Create database... 0.557 seconds
Populate database... 2.020 seconds
Save database... 0.124 seconds
Reload database... 0.049 seconds
Search database... 1.991 seconds
Sort database... 2.106 seconds
Group database... 1.812 seconds
Score: 2483

JRMark (version 14.0.83): 2867

Matt · « **Reply #12 on:** October 14, 2009, 05:59:40 pm »

There's some interesting stuff comparing a Core Duo and i5 / i7.

For example, single-threaded integer math runs about the same speed, but the i7 walks away in the multi-threaded math tests. This isn't too surprising since it has twice the cores.

Single-threaded large scale memory manipulation is about 4x faster on the i7 (for example, Image > Flood filling). I didn't expect that, as I'd always read the jump to DDR3 wasn't exciting.

This also shows that some core database routines still have room to grow to better utilize 4+ CPU cores.

bytestar · « **Reply #13 on:** October 14, 2009, 06:01:16 pm »

Here's my HTPC (Core 2 Duo E6750 @ 3.2GHz, 8GB DDR2-RAM, NVIDIA GT 8600 GPU) Testing on Windows Vista SP2 X64

=== Running Benchmarks (please do not interrupt) ===

Running 'Math' benchmark...
   Single-threaded integer math... 3,579 seconds
   Single-threaded floating point math... 2,226 seconds
   Multi-threaded integer math... 3,683 seconds
   Multi-threaded mixed math... 2,299 seconds
Score: 1612

Running 'Bilder' benchmark...
   Image creation / destruction... 1,388 seconds
   Flood filling... 1,907 seconds
   Direct copying... 2,027 seconds
   Small renders... 3,001 seconds
   Bilinear rendering... 2,869 seconds
   Bicubic rendering... 2,875 seconds
Score: 1564

Running 'Datenbank' benchmark...
   Create database... 1,503 seconds
   Populate database... 4,091 seconds
   Save database... 0,324 seconds
   Reload database... 0,087 seconds
   Search database... 3,035 seconds
   Sort database... 3,055 seconds
   Group database... 2,692 seconds
Score: 1454

JRMark (version 14.0.83): 1543

glynor · « **Reply #14 on:** October 14, 2009, 06:05:05 pm »

Quote from: Matt on October 14, 2009, 05:59:40 pm

There's some interesting stuff comparing a Core Duo and i5 / i7.

For example, single-threaded integer math runs about the same speed, but the i7 walks away in the multi-threaded math tests. This isn't too surprising since it has twice the cores.

Single-threaded large scale memory manipulation is about 4x faster on the i7 (for example, Image > Flood filling). I didn't expect that, as I'd always read the jump to DDR3 wasn't exciting.

This also shows that some core database routines still have room to grow to better utilize 4+ CPU cores.

If you look at my two scores, it is even more interesting to compare the two, since my Core 2 is a quad core that isn't cache-crippled (it has the full 12MB), and it is compared to the Core i5 (which is quad core with NO hyperthreading).

Even still, the Nehalem absolutely dominates the entire Image processing test. Granted, my Core i5 has a 14% clock speed advantage, but look at those Flood Filling numbers. That's far more than the 33% or so IPC advantage the Nehalem core has over the Core 2 derivatives.

gappie · « **Reply #15 on:** October 14, 2009, 06:05:18 pm »

this is my test and crap machine, amd athlon dual core 3800 2.00 Ghz (the first was a intel 2 cpu 6300 1,86 Ghz)
=== Running Benchmarks (please do not interrupt) ===

Running 'Math' benchmark...
   Single-threaded integer math... 11,644 seconds
   Single-threaded floating point math... 4,790 seconds
   Multi-threaded integer math... 11,715 seconds
   Multi-threaded mixed math... 4,840 seconds
Score: 576

Running 'Image' benchmark...
   Image creation / destruction... 2,387 seconds
   Flood filling... 6,125 seconds
   Direct copying... 6,094 seconds
   Small renders... 7,885 seconds
   Bilinear rendering... 7,366 seconds
   Bicubic rendering... 6,085 seconds
Score: 612

Running 'Database' benchmark...
   Create database... 2,372 seconds
   Populate database... 7,658 seconds
   Save database... 0,704 seconds
   Reload database... 0,721 seconds
   Search database... 7,456 seconds
   Sort database... 5,972 seconds
   Group database... 5,074 seconds
Score: 718

JRMark (version 14.0.83): 635

JimH · « **Reply #16 on:** October 14, 2009, 06:07:15 pm »

I think we should sell motherboards and CPU's at the bottom of the benchmark.

glynor · « **Reply #17 on:** October 14, 2009, 06:10:21 pm »

Quote from: JimH on October 14, 2009, 06:07:15 pm

I think we should sell motherboards and CPU's at the bottom of the benchmark.

The margins on hardware are RAZOR THIN. Just have a direct clickthrough to the Newegg ad on Interact instead.

Mr ChriZ · « **Reply #18 on:** October 14, 2009, 06:13:21 pm »

Athlon X2 5000 2GB. ATI Something or other. Plenty of Dust
=== Running Benchmarks (please do not interrupt) ===

Running 'Math' benchmark...
   Single-threaded integer math... 9.204 seconds
   Single-threaded floating point math... 3.835 seconds
   Multi-threaded integer math... 8.995 seconds
   Multi-threaded mixed math... 3.732 seconds
Score: 737

Running 'Image' benchmark...
   Image creation / destruction... 1.774 seconds
   Flood filling... 4.283 seconds
   Direct copying... 3.491 seconds
   Small renders... 6.431 seconds
   Bilinear rendering... 5.627 seconds
   Bicubic rendering... 4.544 seconds
Score: 841

Running 'Database' benchmark...
   Create database... 1.905 seconds
   Populate database... 5.509 seconds
   Save database... 0.595 seconds
   Reload database... 0.433 seconds
   Search database... 5.767 seconds
   Sort database... 4.785 seconds
   Group database... 3.858 seconds
Score: 941

JRMark (version 14.0.83): 840

Like me it appears to be bad at Math.

rick.ca · « **Reply #19 on:** October 14, 2009, 06:14:02 pm »

For comic relief, I offer this...

=== Running Benchmarks (please do not interrupt) ===

Running 'Math' benchmark...
Single-threaded integer math... 21.236 seconds
Single-threaded floating point math... 9.972 seconds
Multi-threaded integer math... 42.180 seconds
Multi-threaded mixed math... 19.590 seconds
Score: 204

Running 'Image' benchmark...
Image creation / destruction... 5.484 seconds
Flood filling... 6.953 seconds
Direct copying... 9.894 seconds
Small renders... 15.124 seconds
Bilinear rendering... 21.788 seconds
Bicubic rendering... 27.864 seconds
Score: 253

Running 'Database' benchmark...
Create database... 7.024 seconds
Populate database... 20.526 seconds
Save database... 0.950 seconds
Reload database... 0.896 seconds
Search database... 11.199 seconds
Sort database... 11.514 seconds
Group database... 12.207 seconds
Score: 334

JRMark (version 14.0.83): 264

I wouldn't dare post post this unless I had a Core i7-860 on order.

Sorry gappie, I get the "crappiest machine" award!

bytestar · « **Reply #20 on:** October 14, 2009, 06:15:51 pm »

Same Hardware but not Windows Vista now testing on Windows 7 X64 RTM.

=== Running Benchmarks (please do not interrupt) ===

Running 'Math' benchmark...
   Single-threaded integer math... 3,506 seconds
   Single-threaded floating point math... 2,214 seconds
   Multi-threaded integer math... 3,929 seconds
   Multi-threaded mixed math... 2,241 seconds
Score: 1598

Running 'Bilder' benchmark...
   Image creation / destruction... 1,341 seconds
   Flood filling... 1,529 seconds
   Direct copying... 1,892 seconds
   Small renders... 2,899 seconds
   Bilinear rendering... 2,687 seconds
   Bicubic rendering... 2,668 seconds
Score: 1690

Running 'Datenbank' benchmark...
   Create database... 1,459 seconds
   Populate database... 3,916 seconds
   Save database... 0,321 seconds
   Reload database... 0,080 seconds
   Search database... 2,926 seconds
   Sort database... 2,948 seconds
   Group database... 2,607 seconds
Score: 1508

JRMark (version 14.0.83): 1599

Quote from: bytestar on October 14, 2009, 06:01:16 pm

Here's my HTPC (Core 2 Duo E6750 @ 3.2GHz, 8GB DDR2-RAM, NVIDIA GT 8600 GPU) Testing on Windows Vista SP2 X64

=== Running Benchmarks (please do not interrupt) ===

Running 'Math' benchmark...
   Single-threaded integer math... 3,579 seconds
   Single-threaded floating point math... 2,226 seconds
   Multi-threaded integer math... 3,683 seconds
   Multi-threaded mixed math... 2,299 seconds
Score: 1612

Running 'Bilder' benchmark...
   Image creation / destruction... 1,388 seconds
   Flood filling... 1,907 seconds
   Direct copying... 2,027 seconds
   Small renders... 3,001 seconds
   Bilinear rendering... 2,869 seconds
   Bicubic rendering... 2,875 seconds
Score: 1564

Running 'Datenbank' benchmark...
   Create database... 1,503 seconds
   Populate database... 4,091 seconds
   Save database... 0,324 seconds
   Reload database... 0,087 seconds
   Search database... 3,035 seconds
   Sort database... 3,055 seconds
   Group database... 2,692 seconds
Score: 1454

JRMark (version 14.0.83): 1543

glynor · « **Reply #21 on:** October 14, 2009, 06:16:23 pm »

Quote from: rick.ca on October 14, 2009, 06:14:02 pm

I wouldn't dare post post this unless I had a Core i7-860 on order.

Sorry gappie, I get the "crappiest machine" award!

Ha ha! You'll love it. You can OC them to the moon.

What was that score from?

JimH · « **Reply #22 on:** October 14, 2009, 06:16:44 pm »

Nice marks, rick.ca. Let's see if anyone can beat it. I have some PC's around here somewhere....

gappie · « **Reply #23 on:** October 14, 2009, 06:17:32 pm »

Quote from: JimH on October 14, 2009, 06:07:15 pm

I think we should sell motherboards and CPU's at the bottom of the benchmark.

guess so.. the question is what those twainmarks mean. the first machine i posted here can record 24 tracks at 24 bits easely. and has no problem with my 'modest' media database. does it also have influence when its done on a 64 bit system? xp vs w7 etc...

nice and intriguing new little tool..

gab

JimH · « **Reply #24 on:** October 14, 2009, 06:17:54 pm »

Quote from: glynor on October 14, 2009, 06:10:21 pm

The margins on hardware are RAZOR THIN. Just have a direct clickthrough to the Newegg ad on Interact instead.

That's a better idea.

gappie · « **Reply #25 on:** October 14, 2009, 06:19:13 pm »

Quote from: rick.ca on October 14, 2009, 06:14:02 pm

Sorry gappie, I get the "crappiest machine" award!

lol... that is a relief.. but you have a new machine on the way...

bytestar · « **Reply #26 on:** October 14, 2009, 06:20:28 pm »

For me it seems Windows 7 is a litle bit faster than Windows Vista SP2 on the same Hardware.

Quote from: gappie on October 14, 2009, 06:17:32 pm

guess so.. the question is what those twainmarks mean. the first machine i posted here can record 24 tracks at 24 bits easely. and has no problem with my 'modest' media database. does it also have influence when its done on a 64 bit system? xp vs w7 etc...

nice and intriguing new little tool..

gab

glynor · « **Reply #27 on:** October 14, 2009, 06:21:33 pm »

Quote from: gappie on October 14, 2009, 06:17:32 pm

guess so.. the question is what those twainmarks mean.

It's a measure of your self-worth, of course, just like ANY benchmark. Silly Gappie...

JimH · « **Reply #28 on:** October 14, 2009, 06:22:12 pm »

Quote from: glynor on October 14, 2009, 06:21:33 pm

It's a measure of your self-worth, of course, just like ANY benchmark. Silly Gappie...

It's an inverse measure of your net worth.

gappie · « **Reply #29 on:** October 14, 2009, 06:23:22 pm »

Quote from: glynor on October 14, 2009, 06:21:33 pm

It's a measure of your self-worth, of course, just like ANY benchmark. Silly Gappie...

lol..

this together with my old volvo... maybe i should read more spam..

glynor · « **Reply #30 on:** October 14, 2009, 06:25:00 pm »

Quote from: JimH on October 14, 2009, 06:22:12 pm

It's an inverse measure of your net worth.

Ha ha!! That too, of course.

Maybe you guys should do a contest where the person with the highest submitted score gets a set of virtual "truck nuts" that hangs off the bottom of their avatar picture on Interact. That'd be sweet.

JimH · « **Reply #31 on:** October 14, 2009, 06:29:00 pm »

Quote from: glynor on October 14, 2009, 06:25:00 pm

Ha ha!! That too, of course.

Maybe you guys should do a contest where the person with the highest submitted score gets a set of virtual "truck nuts" that hangs off the bottom of their avatar picture on Interact. That'd be sweet.

This is slowly spinning out of control .... I love it. Thanks for a good laugh. Several, in fact.

I think a prize for the best, maybe a pair of furry dice for the car. And another for the worst, maybe a new motherboard.

leezer3 · « **Reply #32 on:** October 14, 2009, 06:33:33 pm »

Quote from: JimH on October 14, 2009, 06:29:00 pm

This is slowly spinning out of control .... I love it. Thanks for a good laugh. Several, in fact.

I think a prize for the best, maybe a pair of furry dice for the car. And another for the worst, maybe a new motherboard.

I can fish out the 486 (Normally runs Win 3.11/ DOS) if you're going for that

Don't think I've got anything earlier that'll run XP lol.

-Leezer-

JimH · « **Reply #33 on:** October 14, 2009, 06:33:42 pm »

I had to look up truck nuts. I added a link to glynor's post. Just shows what he's learned on twitter.

gappie · « **Reply #34 on:** October 14, 2009, 06:35:52 pm »

guess i need a pair for my car, and for my compu..

glynor · « **Reply #35 on:** October 14, 2009, 06:38:41 pm »

This is interesting...

Quote from: Matt on October 14, 2009, 05:47:58 pm

Here's an i7 @ 3.2

Running 'Math' benchmark...
   Multi-threaded integer math..1.210 seconds
   Multi-threaded mixed math... 0.825 seconds
Score: 2225

Running 'Image' benchmark...
   Flood filling... 0.450 seconds
   Direct copying... 0.543 seconds

Quote from: glynor on October 14, 2009, 05:57:10 pm

And here's my server machine in the basement (Core i5 750 @ 4.0GHz, 4GB RAM, AMD HD 3850):

Running 'Math' benchmark...
   Multi-threaded integer math... 1.723 seconds
   Multi-threaded mixed math... 1.008 seconds

Running 'Image' benchmark...
   Flood filling... 0.470 seconds
   Direct copying... 0.676 seconds

I've got you beat substantially on a clockspeed basis, with the same exact CPU architecture, but in these four metrics your i7 still beat my system. I wonder if this is the impact of Hyperthreading or the triple-channel memory (or both)? (I'm assuming that you have a Core i7 9-series CPU, and not one of the newer Lynnfield parts, of course). I'd guess hyperthreading mostly, but the triple-channel memory could be having a tiny bit of an impact here, even though my dual-channel RAM is almost certainly running at a much higher clockspeed than yours is on that i7. The extra memory channel could especially help on those fill-rate limited tests.

Still fascinating...

My i5 blew yours away on the Database tests though. Of course, you could easily overclock that thing (assuming you have a decent heatsink) and beat mine, so it is a bit of a pyrrhic victory.

glynor · « **Reply #36 on:** October 14, 2009, 06:48:44 pm »

Quote from: JimH on October 14, 2009, 06:33:42 pm

I had to look up truck nuts. I added a link to glynor's post. Just shows what he's learned on twitter.

I can't believe you don't know what they are, living in Minnesota! I probably can't go a week commuting to work without seeing someone with them here.

It always seems so self-deprecating to me...

ThoBar · « **Reply #37 on:** October 14, 2009, 07:06:31 pm »

Some interesting results: (on a Win7 x64 laptop)

With Power Saver mode on:

Code: [Select]

=== Running Benchmarks (please do not interrupt) ===

Running 'Math' benchmark...
    Single-threaded integer math... 13.814 seconds
    Single-threaded floating point math... 8.929 seconds
    Multi-threaded integer math... 4.817 seconds
    Multi-threaded mixed math... 2.543 seconds
Score: 631

Running 'Image' benchmark...
    Image creation / destruction... 3.987 seconds
    Flood filling... 4.274 seconds
    Direct copying... 2.670 seconds
    Small renders... 6.515 seconds
    Bilinear rendering... 3.249 seconds
    Bicubic rendering... 2.939 seconds
Score: 931

Running 'Database' benchmark...
    Create database... 0.926 seconds
    Populate database... 9.958 seconds
    Save database... 0.572 seconds
    Reload database... 0.251 seconds
    Search database... 7.151 seconds
    Sort database... 8.725 seconds
    Group database... 6.983 seconds
Score: 622

JRMark (version 14.0.83): 728

With High Performance mode on:

Code: [Select]

=== Running Benchmarks (please do not interrupt) ===

Running 'Math' benchmark...
    Single-threaded integer math... 4.257 seconds
    Single-threaded floating point math... 2.522 seconds
    Multi-threaded integer math... 4.299 seconds
    Multi-threaded mixed math... 2.535 seconds
Score: 1396

Running 'Image' benchmark...
    Image creation / destruction... 1.578 seconds
    Flood filling... 2.557 seconds
    Direct copying... 2.519 seconds
    Small renders... 3.069 seconds
    Bilinear rendering... 2.916 seconds
    Bicubic rendering... 2.905 seconds
Score: 1415

Running 'Database' benchmark...
    Create database... 0.829 seconds
    Populate database... 3.339 seconds
    Save database... 0.166 seconds
    Reload database... 0.078 seconds
    Search database... 3.341 seconds
    Sort database... 3.181 seconds
    Group database... 2.844 seconds
Score: 1560

JRMark (version 14.0.83): 1457

I actually would have expected this machine to do a little better, but nevermind $:-\$

Is it strange that the single vs multi-thread performance is so similar?

rick.ca · « **Reply #38 on:** October 14, 2009, 07:08:31 pm »

Quote from: glynor on October 14, 2009, 06:16:23 pm

Ha ha! You'll love it. You can OC them to the moon.

What was that score from?

No point keeping it a secret now... it's from a P4 1.8 MHz with 512 Mb RAM. Needless to say, I'll love the new one without OC'ing. At least I hope so. Not being the truck nuts sort, I decided to go with the lower power supply and smaller case the lower power requirement of the Lynnfield processor seems to allow. I know next to nothing about OC'ing, but I assume that decision might put some constraints on what I can do. I read a comforting opinion that the Turbo Boost feature lessens the need for/benefit from OC'ing.

Matt · « **Reply #39 on:** October 14, 2009, 07:37:31 pm »

Quote from: confishy on October 14, 2009, 07:06:31 pm

Is it strange that the single vs multi-thread performance is so similar?

It's doing twice as much work in the multi-threaded math loops, so a dual-core CPU will have about the same times for all the math stages.

If you look at a single-core CPU, it takes about twice as long for it to do the multi-threaded tests:
http://yabb.jriver.com/interact/index.php?topic=54396.msg370190#msg370190

glynor · « **Reply #40 on:** October 14, 2009, 07:52:48 pm »

Quote from: rick.ca on October 14, 2009, 07:08:31 pm

No point keeping it a secret now... it's from a P4 1.8 MHz with 512 Mb RAM. Needless to say, I'll love the new one without OC'ing. At least I hope so. Not being the truck nuts sort, I decided to go with the lower power supply and smaller case the lower power requirement of the Lynnfield processor seems to allow. I know next to nothing about OC'ing, but I assume that decision might put some constraints on what I can do. I read a comforting opinion that the Turbo Boost feature lessens the need for/benefit from OC'ing.

That's very true. The dynamic core clocking feature is fantastic, and you probably really won't see the need. I run video transcoding for my TV recordings on that machine all the time (automated), so having the higher clockspeed really helps there.

datdude · « **Reply #41 on:** October 14, 2009, 10:59:22 pm »

=== Running Benchmarks (please do not interrupt) ===

Running 'Math' benchmark...
Single-threaded integer math... 4.765 seconds
Single-threaded floating point math... 2.688 seconds
Multi-threaded integer math... 1.241 seconds
Multi-threaded mixed math... 0.941 seconds
Score: 1972

Running 'Image' benchmark...
Image creation / destruction... 0.983 seconds
Flood filling... 0.667 seconds
Direct copying... 0.745 seconds
Small renders... 2.978 seconds
Bilinear rendering... 1.460 seconds
Bicubic rendering... 1.405 seconds
Score: 2671

Running 'Database' benchmark...
Create database... 1.229 seconds
Populate database... 2.841 seconds
Save database... 0.147 seconds
Reload database... 0.061 seconds
Search database... 2.783 seconds
Sort database... 3.523 seconds
Group database... 2.632 seconds
Score: 1627

JRMark (version 14.0.83): 2090

i7 920 2.67 ghz 6gb Ram vista 64 bit

not too shabby but glynor still has us all beat by quite a bit.

I'm wondering what would help these scores besides a better cu? Solid state drive? Faster Ram?

How much of a perceptual difference in MC would that make?

Yaobing · « **Reply #42 on:** October 14, 2009, 11:19:41 pm »

Quote from: JimH on October 14, 2009, 06:16:44 pm

Nice marks, rick.ca. Let's see if anyone can beat it. I have some PC's around here somewhere....

Definitely not beating rick.ca's, but this qualifies for second place. I have not tried my old laptops though, one of which probably would not even run MC14 as it has Windows 98SE on it (I am using it as a printer server, and it works horribly at that too).

=== Running Benchmarks (please do not interrupt) ===

Running 'Math' benchmark...
   Single-threaded integer math... 13.627 seconds
   Single-threaded floating point math... 8.117 seconds
   Multi-threaded integer math... 14.164 seconds
   Multi-threaded mixed math... 8.063 seconds
Score: 432

Running 'Image' benchmark...
   Image creation / destruction... 3.632 seconds
   Flood filling... 4.062 seconds
   Direct copying... 3.710 seconds
   Small renders... 7.388 seconds
   Bilinear rendering... 6.993 seconds
   Bicubic rendering... 6.221 seconds
Score: 687

Running 'Database' benchmark...
   Create database... 2.047 seconds
   Populate database... 8.494 seconds
   Save database... 0.786 seconds
   Reload database... 0.928 seconds
   Search database... 8.559 seconds
   Sort database... 5.510 seconds
   Group database... 5.182 seconds
Score: 682

JRMark (version 14.0.83): 601

Unlike me it appears to be bad at Math. (this line is borrowed from Mr ChriZ)

rick.ca · « **Reply #43 on:** October 15, 2009, 12:12:25 am »

Quote from: Yaobing on October 14, 2009, 11:19:41 pm

Definitely not beating rick.ca's, but this qualify for second place. I have not tried my old laptops though, one of which probably would not even run MC14 as it has Windows 98SE on it (I am using it as a printer server, and it works horribly at that too).

If this is going to be a serious competition for the "floormark," I must insist on some rules. Like the subject machine must be the used to run MC on a daily basis. Compared to some other software, MC actually runs reasonably well on this thing—something you guys should be proud of. Maybe I should send it to you so you can have it bronzed.

jmone · « **Reply #44 on:** October 15, 2009, 03:12:37 am »

Bloody hell what is Glynor Running (or OCing)!

Mine - i7 920 2.67 ghz 6gb Ram vista 64 bit, Old 7600gs, Old WD500GB HDD

=== Running Benchmarks (please do not interrupt) ===

Running 'Math' benchmark...
Single-threaded integer math... 4.748 seconds
Single-threaded floating point math... 2.680 seconds
Multi-threaded integer math... 1.361 seconds
Multi-threaded mixed math... 0.928 seconds
Score: 1955

Running 'Image' benchmark...
Image creation / destruction... 1.063 seconds
Flood filling... 0.679 seconds
Direct copying... 0.791 seconds
Small renders... 2.969 seconds
Bilinear rendering... 1.524 seconds
Bicubic rendering... 1.454 seconds
Score: 2594

Running 'Database' benchmark...
Create database... 1.077 seconds
Populate database... 3.464 seconds
Save database... 0.279 seconds
Reload database... 0.266 seconds
Search database... 3.447 seconds
Sort database... 3.684 seconds
Group database... 2.581 seconds
Score: 1453

JRMark (version 14.0.83): 2001

raldo · « **Reply #45 on:** October 15, 2009, 03:42:09 am »

Quote from: rick.ca on October 14, 2009, 06:14:02 pm

JRMark (version 14.0.83): 264

Ha ha, it's even worse than I thought! (ref. our earlier PvdImport discussions

)

rick.ca · « **Reply #46 on:** October 15, 2009, 06:31:06 am »

Go ahead and laugh. Had it not been for PvdImport, this machine might have lasted another few years.

Listener · « **Reply #47 on:** October 15, 2009, 10:50:10 am »

There may be a use for a PC even slower than rick's P4 running MC.

I have considered using a netbook or a ULV laptop to run MC and play music in one of my rooms. It would be of interest to see how MC works on such a PC retrieving files across a LAN or using a connection to a library server.

If you are using MC on an Atom based PC, post your benchmark results and comment on how MC works on that PC.

Bill

Listener · « **Reply #48 on:** October 15, 2009, 10:55:05 am »

HP desktop - AMD dual Core Athlon 64 X2 dual core cpu at 2.4 GHz,
2 GB Ram, 3 7200 RPM 3.5" SATA drives

--- results
Running 'Math' benchmark...
Single-threaded integer math... 9.592 seconds
Single-threaded floating point math... 3.962 seconds
Multi-threaded integer math... 9.621 seconds
Multi-threaded mixed math... 4.025 seconds
Score: 699

Running 'Image' benchmark...
Image creation / destruction... 1.893 seconds
Flood filling... 3.161 seconds
Direct copying... 3.039 seconds
Small renders... 6.424 seconds
Bilinear rendering... 5.768 seconds
Bicubic rendering... 4.779 seconds
Score: 878

Running 'Database' benchmark...
Create database... 1.226 seconds
Populate database... 5.088 seconds
Save database... 0.305 seconds
Reload database... 0.138 seconds
Search database... 5.685 seconds
Sort database... 5.108 seconds
Group database... 4.096 seconds
Score: 993

JRMark (version 14.0.83): 857

glynor · « **Reply #49 on:** October 15, 2009, 12:26:57 pm »

Quote from: jmone on October 15, 2009, 03:12:37 am

Bloody hell what is Glynor Running (or OCing)!

I referenced it in my original post, but since you asked... The machine that scored the JRMark 2867 score was:

CPU: Intel Core i5 750 overclocked to 4.0GHz (Base Clock 200 MHz x 20 Multiplier with Turbo disabled)
RAM: G.Skill Ripjaws 4GB (2x2GB) DDR3 @ 1600MHz (7-8-7-24 timings)
GPU: AMD HD 3850 @ stock (512MB)
Motherboard: Gigabyte GA-P55-UD4P
(And a whole BUNCH of hard drives, though I don't think this benchmark stresses the storage subsystem too much.)

It probably would have been even higher had I shut down all my background tasks, but I didn't. I couldn't actually, since the system was actually even BUSY at the time recording a TV show via BTV.

INTERACT FORUM

Author Topic: JRMark -- JRiver's Benchmark Tool (Read 239398 times)