A-Data PC4000 at 301MHz(602MHz DDR)...

hipro5 · Sep 20, 2003

How's that???......A friend of mine came into my lab to modd his Mobo, 9800 and ANTEC PSU and he had 2 x 256MB A-Data PC4000 with him........Well I couldn't NOT to test them so I did.........

Mobo : ABIT IC7-G
Ram : 2x 256MB A-Data PC4000 at Cas 2.5 - 7 - 4 - 4 (not very bad) at 3.35Vdimm
Cooling : Modded Prometeia with R404 gas

There you go........

Top Hat Theater · Sep 20, 2003

Leapin' lingosta! That's a scary amount of bandwidth...AMD or Intel? j/k

~THT

hipro5 · Sep 20, 2003

Top Hat Theater said:
Leapin' lingosta! That's a scary amount of bandwidth...AMD or Intel? j/k

~THT

whatever2003 · Sep 20, 2003

One more slap across the face folks in Abit that could not figure out they need to let us adjust voltage in BIOS on our own.
Scary bandwidth!

Kunaak · Sep 21, 2003

how about some Geil 4200 there Hipro?

only 3.275 volts, my 3.3 volt line sensor is holding my Vdimm back...

I will change that soon

Overclocker456 · Sep 21, 2003

Very nice, but I rather be at 2/2/2/5 with something like around 470MHz or so.

Top Hat Theater · Sep 21, 2003

I'd take whatever offered the highest bandwidth and I think running at 604 vs. 470, the 604 would do better.

~THT

hipro5 · Sep 21, 2003

5:4 with aggressive timings is better than 1:1 with lose timings.......

This is a 5:4 bench with EPOX Mobo.....If I had an ASUS one I think l could do better......

Overclocker456 · Sep 21, 2003

that's what I'm talking about.. very nice hipro.

Just curious.. in 3dmark2001SE what do you get at 640x480 16bit?

Lithan · Sep 21, 2003

Strange... You are running at 13 more FSB, and his score is better, yet you say your method is better?

larva · Sep 22, 2003

Lithan said:
Strange... You are running at 13 more FSB, and his score is better, yet you say your method is better?

A couple of salient points come to mind:

1) Unless I missed something only Hipro posted any scores.

2) Sandra buffered bandwidth is next to useless for any kind of performance evaluation. The 3DMark2001 and PCMark 2002 memory score will bear out the truth of what Hipro is saying. I maintain that 1:1 with PAT and 2-2-2-5 timings is actually better than both, but you have to have a higher multiplier than most have to make it come off.

Hipro, have you found the highest fsb you can do 1:1, 2-2-2-5, and PAT at and cranked up the multiplier on your ES chip to match? I'd be curious to know if it doesn't provide surprising results.

hipro5 · Sep 22, 2003

3DMark at 640 x 480 ? ......I haven't tested yet........BUT at the EXACT same CPU frequency(MHz - fsb) with Ram 5:4 and aggressive timings , I get better score in 3DMark than 1:1 and these timings......

hipro5 · Sep 22, 2003

larva you are right....The 1:1 IS a bit better than 5:4 BUT ONLY if both are with aggressive timings.......The max I could bench was 261MHz with x 16 multi and Cas 2-5-2-2 - Street Racer - Enhanced - 5 Clocks - Enable(on IC7-G)......The best of all IS 1:1 at aggressive timings......

larva · Sep 22, 2003

Heh, I thought that would prove to be the case. One aspect that people routinely ignore when claiming 5:4 gives up nothing is the loss in PAT efficiency that occurs at ratios other than 1:1. Intel did a tremendous job optimizing performance at 1:1 through their typical excellent chipset design and the super-agressive effective timings that PAT invokes at the chipset level. This fact flies right over Sandra's head.

My machine's strength (well, for junk) while running the ram at a mere 380MHz shows the potential of the 1:1 mode if accompanied by GAT and/or PAT and 2-2-2-5 timings. It's a lot easier to make happen at 190fsb than 260 though, given today's ram quality.

The 5:4 mode is useful, but it doesn't perform quite as well at 1:1. It does allow people to crank up the fsb to achieve high on-chip clock rates though, and this of course is the most important quantity to be maximized. But to do so is just a bandaid for a too-low multiplier, and those that have the ability to choose a chip with a higher one or change it can reap the benefits of Intel's excellent 1:1 optimizations and produce better overall results.

whatever2003 · Sep 22, 2003

1) Unless I missed something only Hipro posted any scores.

Yes, you missed the fact that the first score he posted isn't on his machine. That score shows that relaxed timings at 1:1 beats his machine that runs 5:4 with tight timmings at 13 MHz higher FSB. Though of course he did not post the clock of CPU there.
How about that going against your thesis? Not that I am advocating any though.

micamica1217 · Sep 22, 2003

whatever2003 said:
Yes, you missed the fact that the first score he posted isn't on his machine. That score shows that relaxed timings at 1:1 beats his machine that runs 5:4 with tight timmings at 13 MHz higher FSB. Though of course he did not post the clock of CPU there.
How about that going against your thesis? Not that I am advocating any though.

unless I'm wrong, you miss understood hipro5....

the moded IC7 "is" hipro5's mobo as well as the epox.

only the sticks are not his.
also, the epox stinks at bandwith scores as compared to the IC7.

mica

larva · Sep 22, 2003

whatever2003 said:
Yes, you missed the fact that the first score he posted isn't on his machine. That score shows that relaxed timings at 1:1 beats his machine that runs 5:4 with tight timmings at 13 MHz higher FSB. Though of course he did not post the clock of CPU there.
How about that going against your thesis? Not that I am advocating any though.

People were wanting to claim that someone else's results disproved what hipro was saying. Again, for those in the back row, only hipro posted any scores.

Secondly, Sandra is no judge of this. I could care less if anyone's scores, hipro's or otherwise, show slightly higher on the 5:4 settings. You might as well just guess which one is faster, Sandra is useless for this comparison.

Overclocker456 · Sep 22, 2003

hipro5 said:
3DMark at 640 x 480 ? ......I haven't tested yet........BUT at the EXACT same CPU frequency(MHz - fsb) with Ram 5:4 and aggressive timings , I get better score in 3DMark than 1:1 and these timings......

yeah I'm sure you'd get a amazing score.. I'm just curious as to how high it would be..

whatever2003 · Sep 22, 2003

People were wanting to claim that someone else's results disproved what hipro was saying. Again, for those in the back row, only hipro posted any scores.

You still don't understand what was said.

larva · Sep 22, 2003

Lithan said:
Strange... You are running at 13 more FSB, and his score is better, yet you say your method is better?

Although it is apparant I can't make this clear for you, the 'his' in the above was referring to someone besides hipro. Hipro posted the only scores in this thread. That was the point I was making.

A-Data PC4000 at 301MHz(602MHz DDR)...

Member

Member

Member

Banned

Member

Member

Member

Member

Member

Disabled

Inactive Moderator

Member

Member

Inactive Moderator

Banned

Member

Inactive Moderator

Member

Banned

Inactive Moderator

Similar threads