Benchmark Reviews Discussion Forum Right Header

Go Back Benchmark Reviews Discussion Forum > Hardware > Overclockers Lounge

Overclockers Lounge Post your projects and results in this section.

Reply
Thread Tools Display Modes
#21
Unread 5th January 2010, 04:09 PM
Olin Coles's Avatar
Olin Coles Olin Coles is offline
Executive Editor
Join Date: Feb 2007
Posts: 2,951
I know this questions compromises your benchmark, but is there any way you could rig a 'short' test? If this benchmark is to be used in real-world testing, it's best to see it finish inside of ten minutes or less. Keep in mind that the benchmark tests that we run for our reviews usually number 3-7 per item per setting. Running this benchmark more than once would really halt progress on our level.
__________________
You can follow Benchmark Reviews on Facebook and Twitter!
Reply With Quote
#22
Unread 5th January 2010, 05:28 PM
XJnine XJnine is offline
Gigabyte
Join Date: Mar 2007
Location: Jacksonville, FL
Posts: 166
Send a message via ICQ to XJnine
Just emailed you the report from my overclocked run. It was faster but not quite as fast as we hoped. :-(
__________________
Asrock Z68 Fatal1ty mobo
Intel Core i7 2600k CPU @ 4.8ghz
Noctua NH-D14 HSF
2x PNY GeForce 680 GTX SLI
32gb G.Skill Sniper DDR3 1866 @ 2133
2x Samsung 840Pro 256gb SSD's, Raid 0
2x Seagate Barracuda 3tb HD's, RAID 1
Vertex 4 128gb SRT Cache Drive
Memorex MRX-800 Blu-Ray drive
Corsair 1000 watt power supply
Fractal Design Define XL R2 Case
Dell U3011 30" monitor
Reply With Quote
#23
Unread 5th January 2010, 09:02 PM
LiquidNitrogenOverclocking LiquidNitrogenOverclocking is offline
Kilobyte
Join Date: Jan 2010
Posts: 35
Quote:
Originally Posted by XJnine View Post
Just emailed you the report from my overclocked run. It was faster but not quite as fast as we hoped. :-(
Dude,

Your test took 4 hours 11 minutes, mine took 8 hours 57 minutes! Your cruising more than twice as fast! I call we have to trade systems.
Reply With Quote
#24
Unread 5th January 2010, 09:06 PM
LiquidNitrogenOverclocking LiquidNitrogenOverclocking is offline
Kilobyte
Join Date: Jan 2010
Posts: 35
Quote:
Originally Posted by Olin Coles View Post
I know this questions compromises your benchmark, but is there any way you could rig a 'short' test? If this benchmark is to be used in real-world testing, it's best to see it finish inside of ten minutes or less. Keep in mind that the benchmark tests that we run for our reviews usually number 3-7 per item per setting. Running this benchmark more than once would really halt progress on our level.
Hi Olin,

So you want a scaled-down version that would finish in about 600 seconds on a top-of-the-line system this is not overclocked?

That would give you roughly 9 orders of magnitude in base 2, meaning after technology doubles the speed for the 9th time, the results will become "decimal places" only.

I could always have it do just the slices with only kings up to 6 pieces, that might not take too long.

And what would this be used for? Like "everybody runs it" on the forum, and it becomes some kind of "standard"?

Let me know, and I can make the coding changes.

Thanks!
Reply With Quote
#25
Unread 5th January 2010, 10:14 PM
ArocOverclock ArocOverclock is offline
Bit
Join Date: Jan 2010
Posts: 2
This is probably just a waste of thread space, but since I have to wait until tomorrow to run my Opteron, I have 2 Intel P4s i decided to dust off, one in a dell dimension 2400 (lol) and another in a computer I built in 2004....

The Dells P4 is just the no frills 2.4GHz Prescott with no bells n' whistles. Even better 768MB of RAM is all shes pushing. the only thing thats not stock in this turd (yeah, i'm using it now...) is the foxconn Mobo that I replaced the original one with.

The other P4 is OC'd @3.0ghz, it's old but it was one of my first stable overclock attempts that I'm surprised still works. It's basically the same Prescott processor thats in the Dell, hooked up to I believe the model was a p4p800 ASUS motherboard with 1Gb or RAM and a GeForce 6800 vid card. The tower I have it in was from an old server.I took a dremel to the access panel, riveted in a piece of Plexiglas and put a 120mm fan in front of the processor and on the top of the case as well .....

sorry for rambling on, extremely tired....anyway the point of all of this is I'm amazed by some of these times being posted and i just want to see how good (or terrible) these p4s will do. I'll post the results when its finished...
Reply With Quote
#26
Unread 6th January 2010, 06:57 AM
Olin Coles's Avatar
Olin Coles Olin Coles is offline
Executive Editor
Join Date: Feb 2007
Posts: 2,951
I'm suggesting a benchmark that completes in ten minutes on a top-end system, so that it might be used in processor/memory/motherboard reviews. I'm leaving for CES this morning, so I'll have to test this in about a week.
__________________
You can follow Benchmark Reviews on Facebook and Twitter!
Reply With Quote
#27
Unread 6th January 2010, 08:44 AM
Olle P Olle P is offline
Terabyte
Join Date: Nov 2008
Location: Sweden
Posts: 455
If I've understood it correctly this benchmark will pretty much boil down to testing the single thread performance of the CPU and very little else in the system, right?
That should make a high paced single or dual core CPU perform better than a slow paced quad core CPU (not accounting for the impact of more level 3 cache in the quad)...

I expect my computer would complete this test in about 10 hours, since it's an older 2.4 GHz Athlon64 X2...

Cheers
Olle
Reply With Quote
#28
Unread 6th January 2010, 08:52 AM
LiquidNitrogenOverclocking LiquidNitrogenOverclocking is offline
Kilobyte
Join Date: Jan 2010
Posts: 35
Quote:
Originally Posted by Olin Coles View Post
I'm suggesting a benchmark that completes in ten minutes on a top-end system, so that it might be used in processor/memory/motherboard reviews. I'm leaving for CES this morning, so I'll have to test this in about a week.
OK, I'll have to give this some thought. Solving the 2-piece through 5-piece databases plus solving the 3 kings vs. 3 kings slice of the 6-piece database should take almost 5 minutes, so doing it twice would be 10 minutes.

The 3 kings vs. 2 kings + 1 checker slice of the 6-piece database takes 45 minutes, so this would not be good.

I also have been giving some thought as to why XJNine did not get an "exact" 33% speedup, as we would expect when going from a 3.0 GHz system to a 4.05 GHz system.
  1. Perhaps his 4.05 GHz estimate is for a "peek" condition, and the average performance for the duration is slightly less than this.
  2. On the larger database slices, such as 2 kings + 1 checker vs. 1 king + 2 checkers, there is much more move generation at the front end of the calculation compared to the back end of the calculation. The program resolves fewer and fewer positions with each pass, so most of the time is spent "parsing the database" than actually "solving" at the end. I would think this would tend to produce a "flatter gain" with increased processing speed, but still, a for-loop is a for-loop is a for-loop, and this must also execute faster on a faster system.
  3. There is no number 3! I can really on think of 2 things!

Last edited by LiquidNitrogenOverclocking : 6th January 2010 at 08:53 AM. Reason: arrrrrg typos
Reply With Quote
#29
Unread 6th January 2010, 08:59 AM
LiquidNitrogenOverclocking LiquidNitrogenOverclocking is offline
Kilobyte
Join Date: Jan 2010
Posts: 35
Quote:
Originally Posted by Olle P View Post
If I've understood it correctly this benchmark will pretty much boil down to testing the single thread performance of the CPU and very little else in the system, right?
That should make a high paced single or dual core CPU perform better than a slow paced quad core CPU (not accounting for the impact of more level 3 cache in the quad)...

I expect my computer would complete this test in about 10 hours, since it's an older 2.4 GHz Athlon64 X2...

Cheers
Olle
Hello Olle,

The slowest time so far at:

http://www.liquidnitrogenoverclockin..._results.shtml

...is just a few minutes under 9 hours, and that was a 2.4 GHz single processor AMD. You would probably be closer to 9 hours than 10 hours, based on my best guess.

You are correct: Single-threaded app, 64-bit, pretty much uses all of the bits during move generator (really good exerciser of the CPU), so a faster single CPU should beat a slower CPU x more cores.
Reply With Quote
#30
Unread 6th January 2010, 01:24 PM
XJnine XJnine is offline
Gigabyte
Join Date: Mar 2007
Location: Jacksonville, FL
Posts: 166
Send a message via ICQ to XJnine
I wonder if my time is related to my memory timings. When I overclocked my CPU my RAM speed went up to 900mhz from 800 at stock speeds but my timings went to 5-5-5-15 from 4-4-4-12. Could it be that the benchmark is more sensitive to latencies than bandwidth of the memory? If the benchmark is not moving the ~900 megs around and just accessing it, it seems like increasing memory latency will negatively impact the benchmark scores, potentially more than the slight increase gained from the 100 mhz increase in RAM speed.

Make sense? I can run it again with my CPU overclocked but at the 800/4-4-4-12 memory settings and see what kind of time I get if you'd like.
__________________
Asrock Z68 Fatal1ty mobo
Intel Core i7 2600k CPU @ 4.8ghz
Noctua NH-D14 HSF
2x PNY GeForce 680 GTX SLI
32gb G.Skill Sniper DDR3 1866 @ 2133
2x Samsung 840Pro 256gb SSD's, Raid 0
2x Seagate Barracuda 3tb HD's, RAID 1
Vertex 4 128gb SRT Cache Drive
Memorex MRX-800 Blu-Ray drive
Corsair 1000 watt power supply
Fractal Design Define XL R2 Case
Dell U3011 30" monitor
Reply With Quote
Sponsored links
Reply


Thread Tools
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

vB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Forum Jump


All times are GMT -7. The time now is 04:33 PM.

Benchmark Reviews Discussion Forum
Powered by vBulletin® Version 3.6.8
Copyright ©2000 - 2017, Jelsoft Enterprises Ltd.
Remove advertisements

Follow Benchmark Reviews on FacebookReceive Tweets from Benchmark Reviews on Twitter