Jump to content

My new system


Recommended Posts

  • Replies 356
  • Created
  • Last Reply

Top Posters In This Topic

well, i'm running 3.0GHz right now on air. it's just that i haven't gotten all the timings figured out so my bandwidth is down a bit.

 

i was able to get 4100mb/s out of my NF2 around 256~260mhz so i know the timings are whats holding my mem bandwith down to around 3790 at 250mhz. i know Trfc made a difference on my NF2 but Trc didn't so much and it's the same on my NF3, i get the same performance wheter it's set to 7 or 11; but yes, Trfc makes a difference but i've found that the two setting play off each other. if one is too high and the other is too low, my performance gets worse, and same goea vice versa. it's all the other timings i need to work with that aren't in the NF2 bios. the Trfc and Trc are old news.

 

TGM

i allway keep mine 3 apart 7-10- 8-11 9-12

Share this post


Link to post
Share on other sites

Yea this avatar is kinda like my personality and I think it's kool too. Thnx AG. GM I have finally settled on my 24/7 clock. Its a bit slower than I had hoped but it games like a banshee. This vid card stutters like hell on lower mem speeds so I cranked them up and lowered the core speed. Overall I'm very happy with this rig and grateful to DFI for making a sckt 754 PCI-E solution for psycho gamers like me that have to get by on a somewhat limited budget. Keep us posted on your progress.

Share this post


Link to post
Share on other sites

  • 2 weeks later...
Yea this avatar is kinda like my personality and I think it's kool too. Thnx AG. GM I have finally settled on my 24/7 clock. Its a bit slower than I had hoped but it games like a banshee. This vid card stutters like hell on lower mem speeds so I cranked them up and lowered the core speed. Overall I'm very happy with this rig and grateful to DFI for making a sckt 754 PCI-E solution for psycho gamers like me that have to get by on a somewhat limited budget. Keep us posted on your progress.
w0w, 3025 with 1.65vcore and you have it water cooled right? with a DD setup?

 

i must have gotten a bit better piece of silicon that you. i'm running, on air 948U heatsink, at 3.0ghz with 1.55vcore idle temp 34C full load temp 46C. i run 10x300 / DDR500 2-3-3-5-1T TCCD.

 

i got my MCP-5 liquid cooling solution from DD's website and i'm in leak testing phase for my water cooling right now. last time i was able to get 3.2ghz and now that i've increased my air cooling overclock, i'll presume i can increase my water cooling overclock. i'll be trying for 3.3ghz 10.5x315 / DDR510 2-3-3-5-1T TCCD.

 

here's a pic of my water cooling in leak testing phase, so far so good. got the one pesky leak stopped. that thing took me about 30mins to get it seated fully.

 

DSCF0021.jpg

 

TGM

Share this post


Link to post
Share on other sites

Very nearly identical at current settings. Max core speed (3.2G) yields a 28 and change 1M. I'd post it but I still haven't spent the time to figure out how.
w0w, 3.2ghz gets you 28 and change ....hmmm, i wonder what 3.2ghz will get me if 3ghz gets me 29 ...in fact, i'm betting i'll go 3.3ghz on water.

 

why would yours be a bit slower? what mem bus are you running and timings at 3.2ghz?

 

TGM

Share this post


Link to post
Share on other sites

Once you get around 39-43k in everest reads superpi scores don't change at all.

It's all cpu from there.

 

It's that way for me even at 266 with some pretty slack timings considered even by me.

It's just something I finally realized these last few days.

 

Cooling, the cooler the cpu, the faster scores you get in superpi too.

I'm sure there is a drop off point but still.

The hotter the cpu, the higher the score(worse).

 

This is this way even though I don't have this memory controller in the cpu.

So it's all cpu.

Even if you don't have the pic enabled it the cpu it's that way, apic.

Share this post


Link to post
Share on other sites

Once you get around 39-43k in everest reads superpi scores don't change at all.

It's all cpu from there.

 

It's that way for me even at 266 with some pretty slack timings considered even by me.

It's just something I finally realized these last few days.

 

Cooling, the cooler the cpu, the faster scores you get in superpi too.

I'm sure there is a drop off point but still.

The hotter the cpu, the higher the score(worse).

 

This is this way even though I don't have this memory controller in the cpu.

So it's all cpu.

Even if you don't have the pic enabled it the cpu it's that way, apic.

i think your opinion is based more off your nf2/Barton system than our nf3-4/A64 systems. also, the temp of the cpu has no bearing on superPI scores as i've ran both with air cooling and water cooling and the scores stay the same with the same settings, unless you're talking about thermal throttling that doesn't happen till 90C. you are right about system latencies, tho your estimate of latency spread is off a bit imho, as they are indicative of the on die memory controller to core bus latency internal to the core. the only way to lower the Everest latency after overclocking the system memory bus and tightening up their timings, is to overclock the core higher since the internal memory controller runs at the same speed as the core itself. or is this what you were trying to say?

 

so dropping the latency by overclocking the core will effect the superPI score, but so will the increased cpu frequency. and this is after finding your best mem bus with frquency vs. timings.

 

i'm a 29.328s and the only way to get a lower score, for me, from here will be to raise my mem controller+core frequency higher. which i'll be doing if i ever get through all my other mobo testing with my 754 chip i need to do first. :sweat:

 

right now i'm stuck at the point of where the heat is holding me back, on air, no more voltage passed 1.55v will get me anything over 3080mhz. but i'll be going for 3300mhz+ when i get the system all put together with the water cooling. on a side note, i also noticed that running a mem bus of 235mhz at 2-2-2-5-1T gives me the same superPI score as running 255mhz at 2-3-3-5-1T, plus the system feels faster ...but running the cpu down from 3000mhz to 2600mhz really increased the latency in everest.

 

TGM

Share this post


Link to post
Share on other sites

Please sign in to comment

You will be able to leave a comment after signing in



Sign In Now

×
×
  • Create New...