Tight Timings vs High Clock Frequencies article

[unsmart]

I was really looking forward to reading this and finding out where the sweet spot is for timings and speed.
All synthetic benches and all on one system, Wtf. Don't run 3dmarko1 and tell me it's the same as running fear, run fear and show me. Why not run a dual core,single core and a p4 then say it's not worth it. Theres no regard for chipset, cpu or real world results. I want to see gaming, encoding and multi tasking results on a few systems.
5% is a real increase if you ask me, if a vid card had that over the competition they would crown it king and we would pay $200 more. Most of us go to great length for 5% more, hell even 2%.
I think this guy just had to meet a dead line and used his own system. I just want this to be handled like a mobo test or graphic card with more then one settup and some real world apps. Just a waste of ten pages.
I say buy good ram push it as high as it will go at low timings, 3%-5% is better then no increase. Plus good hardware is good hardware.

 

[7H4_D00D3]

Right now I'm doing my own research about the subject, here are my specs & results

ASROCK K8VM800
Sempron 3000+ palermo 1.8Ghz,64Kb+64Kb L1 ,128Kb L2
2x512Mb Corsair ValueRAM DDR modules DDR400 2.5-3-3-8
Western Digital 7200rpm 80Gb SATA

SUPERPI (in seconds, less is better)
200HTT 2.5-3-3-8@2t 52,215
215HTT 2,5-3-3-8@2T 50,109
210HTT 2-3-3-8@1T 48,094
215HTT 2,5-3-3-8@1T 47,375

EVEREST MEMORY READ (Mb/s more is better)
200HTT 2.5-3-3-8@2t 3032
215HTT 2,5-3-3-8@2T 3256
210HTT 2-3-3-8@1T 3182
215HTT 2,5-3-3-8@1T 3268

EVEREST MEMORY WRITE (Mb/s more is better)
200HTT 2.5-3-3-8@2t 1156
215HTT 2,5-3-3-8@2T 1238
210HTT 2-3-3-8@1T 1409
215HTT 2,5-3-3-8@1T 1443


SCIENCEMARK 2.0 32.BIT (more is better in all tests)

MOLECULAR DYNAMICS
200HTT 2.5-3-3-8@2t 691.5
215HTT 2,5-3-3-8@2T 766.65
210HTT 2-3-3-8@1T 734.56
215HTT 2,5-3-3-8@1T 764.15

PRIMORDIA
200HTT 2.5-3-3-8@2t 617.95
215HTT 2,5-3-3-8@2T 679.92
210HTT 2-3-3-8@1T 665.02
215HTT 2,5-3-3-8@1T 686.3

CRYPTOGRAPHY
200HTT 2.5-3-3-8@2t 830.56
215HTT 2,5-3-3-8@2T 910.81
210HTT 2-3-3-8@1T 904.94
215HTT 2,5-3-3-8@1T 927.2

STREAM
200HTT 2.5-3-3-8@2t 706.99
215HTT 2,5-3-3-8@2T 761.9
210HTT 2-3-3-8@1T 831.72
215HTT 2,5-3-3-8@1T 851.78

MEMORY BENCHMARK
200HTT 2.5-3-3-8@2t 1130.17
215HTT 2,5-3-3-8@2T 1215.28
210HTT 2-3-3-8@1T 1341.71
215HTT 2,5-3-3-8@1T 1349.1


BLAS/FLOPs
200HTT 2.5-3-3-8@2t 852.93
215HTT 2,5-3-3-8@2T 929.89
210HTT 2-3-3-8@1T 935.56
215HTT 2,5-3-3-8@1T 946.96


SCIENCEMARK TOTAL SCORE
200HTT 2.5-3-3-8@2t 782.3
215HTT 2,5-3-3-8@2T 855.17
210HTT 2-3-3-8@1T 862.66
215HTT 2,5-3-3-8@1T 885.01


I know this is all synth, but it's what I've done so far, I'll do TMPGEnc trials later

 

[gomerpile]

I find this way to weired I had it out with a dude about this in a post I showed him my benches and clock speeds and he said bs. I could'nt care about the bs part but I showed him I flashed my system bios and video bios. What I did was change the timings for the ram and cpu what I did was slow down the ram timing a bit to increase the clock speed than benched. I than went vise a versa and discoverd that the inbetween setting was perfect and I was able to go 125 mhz faster than the clock speed with tight timing for the ram. That speed was the video the cpu is 385 mhz faster than the clock speed

 

[7H4_D00D3]

Lost my wife, lost my car, lost my home, lost my kids still have my games and computer. My next life I will be MR.Gates.

You didn't solve your addiction issues, did you?

Bout your post, how did you proceed?

 

[gomerpile]

No dude I'm getten even worser I spent my rent this month too on computer junk

 

[7H4_D00D3]

No dude I'm getten even worser I spent my rent this month too on computer junk

http://www.megatokyo.com/index.php?strip_id=11
http://www.megatokyo.com/index.php?strip_id=37
http://www.megatokyo.com/index.php?strip_id=38
http://www.megatokyo.com/index.php?strip_id=41

 

[gomerpile]

:lol: :lol: :lol: :lol: :lol: :lol: :lol: :lol: :lol: :lol: :lol: :lol: :lol: :lol: :lol:

 

[MadModMike]

You could always grab OCZ's Extended Voltage series and get DDR500 @ 2-2-2-5, though it takes somethin' like 3.6v to achieve it.

~~Mad Mod Mike, pimpin' the world 1 rig at a time

 

[MadModMike]

I say buy good ram push it as high as it will go at low timings, 3%-5% is better then no increase. Plus good hardware is good hardware.

3-2-2-7 1T
The performance lies in RAS to CAS delay while the CAS can be easily made up with Mhz.
Works for Crucial Ballistix 512MB PC3200/PC4000 only.

The other more expensive altrnative is to use modules with Winbond BH-5 then push it to 250Mhz+ 2-2-2-7 1T using ~3.4V

From what I could tell, the Row Precharge (x-x-X-x) is the main hitter on performance, than again, if you look at the way memory is accessed, you'd think CAS would be the main one.

~~Mad Mod Mike, pimpin' the world 1 rig at a time