Printable View
Depends if you are on meth when viewing it.... :D I doubt it will be completely open when released but Intel has really listened to the community on this board I believe, will probably have to eat those words now...Quote:
Originally Posted by eva2000
Been playing with me Gigabyte P35C. Although something seems to happen around 417 FSB Everest and Sandra bandwidth dont show big differences. Spi does. All settings from bios:
415FSB
http://img.photobucket.com/albums/v6...oc/415FSB1.jpg
420FSB
http://img.photobucket.com/albums/v6...oc/420FSB1.jpg
Difference is bigger in Sandra and Everest from 347 to 348FSB:
347FSB
http://img.photobucket.com/albums/v6...oc/347FSB1.jpg
348FSB
http://img.photobucket.com/albums/v6...oc/348FSB1.jpg
Tried from 300FSB up to 530 with +10mhz a step and checked with both Everest latency and Read. All "normal" except for screens above.
I still have yet to find a strap change from 300-470mhz FSB on my P965-S3 Rev 1.0... Haven't checked for <300mhz FSB strap change but pretty pointless anyways for me having a 8x multi cpu...
same here loc.o
i just boot @ 416 n use easytune5
is there a higher strap than @ 416?
i will go on testing as soon as get some better ram
beta2 version with Performance Level: memset33beta2
Impressive influence on P35 performance, great work The Stilt :)
Hmm I have tested different settings on a Gigabyte P35-DS3P with bios F5B
Conclusion is, that performance levels do matter.
8x 450 MHz (ddr900) does not gain performance over 9x 400 Mhz (ddr800) in superpi 1M.
Level 7 looks as the tightest level possible at 400/450 MHz.
At higher FSB and performancelevels (7+) the RAS# Precharge time at 2 clockcycles seems te be slower then 3 clockcycles. Could be my memory, or the bearlake chip.
Results:
450x 8 4-3-3-7 level 7 14.109s
450x 8 4-3-2-7 level 7 14.078s
450x 8 4-3-3-7 level 9 14.125s
450x 8 4-3-2-7 level 9 14.140s (?)
450x 8 4-3-3-7 level 11 14.141s
450x 8 4-3-2-7 level 11 14.156a (?)
400x 9 4-3-4-7 level 7 14.109s
400x 9 4-3-3-7 level 7 14.094s
400x 9 4-3-2-7 level 7 14.078s
400x 9 4-3-2-7 level 9 14.125s
400x 9 4-3-2-7 level 11 14.125s
Wow, Awesome find.
Gives me a reason to actually upgrade to bearlake.
nice works Stilt and FELIX, thank you.
Me tested with perf. level 7 and 9. Bandwidth inscrease was much :up:
For those of us who didn't get the memo on what Performance Level does, could you give us a 101 on it?
I have two more questions.
First, is there a reason why MemSet 3.3beta2 doesn't show TWR or is it there but I don't recognize it?
Second, I set TWTR to 11 in the BIOS but the second MemSet 3.3beta2 shows 19 and then I set TWTR to 3 in the BIOS but the third MemSet 3.3beta2 shows 11. Any idea why this is happening?
| -- set in BIOS Auto DDR2-1067 -- | set in BIOS 5-5-5-18 6-42-11-3-3-2T | set in 5-5-5-9 3-30-3-3-6-2T |
Attachment 60685
Attachment 60686
just read the post.Quote:
Originally Posted by msgclb
its pretty basic stuff.
The problem was that I couldn't read but that I failed to comprehend. After reading that first post a couple more times, I believe that I get the gist.Quote:
Originally Posted by Kunaak
Setting Performance Level 7
Attachment 60690
Everest Latency
Performance Level 10: 71.0 ns
-Performance Level 9: 68.9 ns
-Performance Level 8: 66.1 ns
-Performance Level 7: 63.7 ns
-Performance Level 6: locked up
Super PI 32M
Performance Level 10: 14m 27.296s
-Performance Level 9: 14m 19.454s
-Performance Level 8: 14m 14.625s
-Performance Level 7: 14m 10.281s
-Performance Level 6: locked up
so does 1t work well on any of these boards?
this is GREAT!...THANK YOU!:)Quote:
Originally Posted by FELIX
i opened beta 2 and saw performance set at 6 already for the Abit IP35....i tried '5' and it went flying out of windows...LOL:D...rebooted and tried '7' and '8' and then ran sandra memory...those runs had about 100 to 200 points lower in bench results...i then reset 6 and ran sandra bench again and i was up 100 points from my previous run before intalling beta 2!..it was like 8350 and now 8450...SUPER!:)
going to re-run PCMark 2005 to see how it scores with "6" set for sure.
Can we get a bios edit with these setting for the P5K Deluxe?
Great work Stilt, you and FELIX are making a big impact :clap:
great job guys
loco did you try the f3b bios i got a slight increase with it
Thats the bios im using now. New memset works great here thanks a lot !!
I tested Memset 3.3 beta 2 with performance level ... and it gave me some strange result.
1. Good improve with 1:1 and fsb 500
2. Freez when I change PL (Performance Level) from default to tighter with div 4:5
3. Strange boot settings with CAS=5 (PF=6 and freez when change to 5) but PL=5 when CAS=4 !! That's not logical ... Both screen booted with identical sets in bios exept Cas
http://img502.imageshack.us/img502/9141/544gc3.jpg
http://img517.imageshack.us/img517/2190/444tx7.jpg
If I could set PL to 5 than CAS=5 wouldn't be slower than CAS=4 ...
hmm that's really weird, but interesting at the same time. I feel like cas4 is almost easier with my dfi p965 than cas5 at 500fsb, not that it's that related to an asus p35 board...
This fine-tuning of memory performance has been made adjustable in BIOS since its 1st release. Where? Setting is called Transaction Booster with Disabled [Relax - 0,1,2,3] and Enabled [Boost - 0,1], with Relax 3 being the highest latency and Boost 1 as the lowest latency available to be set in BIOS by the user. Additionally, DRAM Static Read Control set to Enabled will also provide a boost, as opposed to Disabled <-- that's why at certain high FSB and memory speed, system will often fail to POST with this last option Enabled. It's happening more often with Asus P5K3D (DDR3) flavor, rather than P5KD, due to high memory frequencies...Quote:
Originally Posted by Warboy
From what I know, this option does not change bus strap but affects memory performance by manipulating internal chipset latencies. These are calculated based on FSB, and memory speed and I think tCL in some cases.
:toast:
Very awesome :) Going to have to try this out tonight!
Hmmm.. now I am very interested in what de Intel 975X will have at performance level :eek:
Maybe the 965's are just 975x's with slower performance levels :rofl: ... And that's why the Intel 965 chips will reach a higher FSB limit.... just thinking.
I don't believe it will work on 975 though.Quote:
Originally Posted by Tutti-frutti
130nm compared to 90nm.
Strange Yotomeczek. Board/bios i guess. I tried and default stays 8 for me board/mem with cas4.
Cas 4
http://img.photobucket.com/albums/v636/loc_oc/cas4.jpg
4/5 PF6
http://img.photobucket.com/albums/v6...5500fsbPF6.jpg