2-2-2-5 at ddr400 or 3-2-3-6 at ddr520?
- Thread starter Gommsta
- Start date
AMD 64 always runs asyncronous even with memory at 1:1. I don't have any experience of such good ram but from what I've read DDR520 with CAS 3 probably won't be enough to beat the DDR400 at CAS 2.
Maybe SuperPI 1M will show you the difference so give that a try on both settings and post back with the results.
Maybe SuperPI 1M will show you the difference so give that a try on both settings and post back with the results.
That's a good result for DDR520 at CAS 3. I wasn't expecting such a good showing in SuperPI.
For gaming performance you could try a benchmark in Doom 3 or another game that relies on high CPU speed and low memory latency.
I'm still relatively new to AMD 64 systems so I can't really say much else but hopefully someone with more knowledge/experience will be able to give some advice on the differences between DDR400/tight timings and DDR500+/loose timings.
For gaming performance you could try a benchmark in Doom 3 or another game that relies on high CPU speed and low memory latency.
I'm still relatively new to AMD 64 systems so I can't really say much else but hopefully someone with more knowledge/experience will be able to give some advice on the differences between DDR400/tight timings and DDR500+/loose timings.
Personally I'd go for the DDR520. While latencies are important on the A64 they are not the be-all and end-all as some would have you believe.
As an example a while back I did some tests (using Quake3 at 640x480 to focus on cpu/memory performance) and found that 2.5-3-3-6(?) DDR470 was faster than 2-2-3-6(?) DDR423. I can't test CAS3 because my Twinmos RAM hates it.
The bottom line it depends on each persons setup and how much extra FSB they can get by relaxing their timings - best thing is to test as you have done and choose the fastest option.
As an example a while back I did some tests (using Quake3 at 640x480 to focus on cpu/memory performance) and found that 2.5-3-3-6(?) DDR470 was faster than 2-2-3-6(?) DDR423. I can't test CAS3 because my Twinmos RAM hates it.
The bottom line it depends on each persons setup and how much extra FSB they can get by relaxing their timings - best thing is to test as you have done and choose the fastest option.
Some games like memoryspeed, others like tight timings but the difference between them is under 10% in almost all situations so it really doesn't matter all that much.
But out of the choices you have I would run at DDR520 as the timings are still quite good, normally RAM needs to run at 3-4-4-8 at >DDR500
But out of the choices you have I would run at DDR520 as the timings are still quite good, normally RAM needs to run at 3-4-4-8 at >DDR500
I have noticed absolutley no difference running xp with 2 gig 3500LL 2-3-2-5 to 2 gig gskill @3-4-4-8. 3dmark06 scores were higher with the gskill running @ above timings @280mhz(1:1) than running the 3500LL's @220 mhz(/166). So I would choose slacker timings with a higher bandwidth.
Oh and the 3500LL's will not run 438 stock, they run @ 400 unless you increase the fsb by 17
. I found the 3500's to be excellent memory but would not clock higher than 446 (223@2-3-2-5) without using a divider and even then it had to be a 166 divider coz they didn't work well with a 180 divider. Even relaxing the timings failed to increase stability. I recommend putting 2.8v through them as this was a sweet spot for my set.
Oh and the 3500LL's will not run 438 stock, they run @ 400 unless you increase the fsb by 17
. I found the 3500's to be excellent memory but would not clock higher than 446 (223@2-3-2-5) without using a divider and even then it had to be a 166 divider coz they didn't work well with a 180 divider. Even relaxing the timings failed to increase stability. I recommend putting 2.8v through them as this was a sweet spot for my set.Ok, well by stock I meant the DDR438 2-3-2-6. (actually did DDR440 and 220x10 on the cpu). Everything else I agree with you although i've had them stable 260 at 3-2-3-6 for over a week (Memtest, Prime, SnM etc).
I'm finding in general that things are marginally quicker using relaxed timings and a higher fsb. Only talking 1 fps if that. Makes you wonder whether its worth spending all that money on quality RAM at all. I guess it depends on what revision you get.
Bit disillusioned but hey ho. Lesson learned!
Cheers everyone
Gommsta
I'm finding in general that things are marginally quicker using relaxed timings and a higher fsb. Only talking 1 fps if that. Makes you wonder whether its worth spending all that money on quality RAM at all. I guess it depends on what revision you get.
Bit disillusioned but hey ho. Lesson learned!
Cheers everyone
Gommsta
Permabanned
- Joined
- 7 Jul 2005
- Posts
- 710
- Location
- Suffolk
i was about to ask a similar question
520 at 3-3-2-5 or 430 at 2.3.2.5
520 at 3-3-2-5 or 430 at 2.3.2.5
Associate
- Joined
- 30 Jun 2006
- Posts
- 1,992
- Location
- Cardiff
So what would be an ideal choice. The 3500LL and overclocking to run at 500mhz or Mushkin Redline XP?
Both are fairly pricey. But I feel they would be an upgrade over my corsair value select. Don;t get me wrong I find the value select fairly good value and performs well at the price point but I am trying to squeeze more speed out of my system and I think my memory is holding me back.
Both are fairly pricey. But I feel they would be an upgrade over my corsair value select. Don;t get me wrong I find the value select fairly good value and performs well at the price point but I am trying to squeeze more speed out of my system and I think my memory is holding me back.