G.Skill DDR500 RAM rocks

[firewolfsm]

i started OCing my proc at 1:1, and got to the rated 250MHz easily, then i took the ram all the way up to 270MHz at the stock 2.6V. maybe i got a lucky part, but i am buying this kind of ram for a while. the high bandwidth actually raised my 3DMark scores a good amount too

 

[TrevorRC]

Did the same thing here.
Take the voltage down to 2.5v, and start drooling!

 

[Diasper]

What timings are you guys using?

 

[Painman]

I'm using the HZ 1GB x 2 (I'm assuming this is what the OP is talking about) on the DFI Ultra-D with the default settings given by the 704-2BTA BIOS from DFI-Street... orange slots... board fired up and has been running w/o a single hiccup. Just running it @ the stock DDR500 settings for now, 3-4-4-8 1T on 2.6 volts. I'm confident I could tighten it up to 3-3-3-7 @ 250 (CAS 2.5 seems very unstable) or get 3-4-4-8 to run @ 275. I tried that briefly, but MEMTest86 Test 5 barfed out 3,000 errors.

I'm not in a hurry to start tweaking and testing though, right now I'd rather enjoy the stability I have

 

[Fallengod]

Are you guys refering to this? LINK

Ive been sorta debating on which pc-4000 2gb kit I should buy. Been considering that or OCZ over at monarch. Not really sure which to get.

 

[Painman]

Originally posted by: modempower
Are you guys refering to this? LINK

Ive been sorta debating on which pc-4000 2gb kit I should buy. Been considering that or OCZ over at monarch. Not really sure which to get.

^^ That's the stuff I'm referring to (and assuming the OP is referring to), yes.

 

[firewolfsm]

ya that's the same RAM, cheap, cool, and high speed. my only problem is that this RAM doesn't do good at lower speeds with high timings, just keep the stock 3-4-4-8 timings and the speed will go higher

 

[OvErHeAtInG]

I got this memory Monday, and finally last night, got it installed and tested. Looks like this is the most general UCCC thread so here goes. When set to SPD, it gives timings of 3-4-4-8, which I guess puts me in a lucky minority here. However, it won't overclock at all past 250MHz with any timings - it gives a few errors right at the end of Memtest86 test 5 even at 255MHz, and quite a few errors at the same place at 260. My mobo only has the option for 2.5v and 2.7v, and I've tried every permutation of voltage, timings, nothing. Like someone said (here or elsewhere) it could be the memory controller itself needs more voltage for 2GB, we'll see, my vcore on this mobo only goes up to 1.4v. But I probably just got a set that won't go over 250.

The good part is that it will run 5/6 divider (224 MHz) 2.5-3-3-8, yielding slightly higher performance that my old mismatched Corsair at 2/3 (176) 2-2-2. That, in the end, was what I was hoping for. I should test to see how high the memory itself will go with 2.5-3-3-8.

I've been priming since last night - Large FFT's on core 1 and blend on core 2. The blend errored out after exactly 8 hours 30 min - that's "stable" for me. Judging by the experiences some were having with this RAM, I had genuine hopes of running 1:1 but it was not to be.

 

[Shimmishim]

what board are you trying this out on?

can you mess with Trp, Twr, Twtr, and Trtw settings?

I've tried 3-4-4-8 @ 260 with 3-2-2-2 which wouldn't even boot up (ran fine with TCCD).

i loosened them to 4-3-2-3 and it booted up fine.

@Painman: 250 3-3-3-7 won't work for me. The best I've seen anyone do is 3-3-4-8 or 3-4-3-8 (one of the two) but they had some awesome sticks.

 

[tomt4535]

my set on my DFI NF4 Ultra Infinity is at 260mhz 3-4-4-8 2.6v 24/7, but ive gotten it to run 275Mhz no problem. My X2 3800+ dosent like it that high though, so I had to step it down a bit. Im certianly happy with both parts though, this system screams.

 

[5t3v0]

I have the 3-5-5-10 spd which wouldn't run stable at a mere 252MHz (8x315 with 166 divider). With the same divider on 9x280 I get 229MHz which I having been running on 3-3-3-6 error free for 2 months. I'm currently Prime95 blend testing on 2.5-3-3-6 which has been stable so far after 1.5 hours. I've posted all my timings in this thread.

 

[OvErHeAtInG]

Originally posted by: Shimmishim
what board are you trying this out on?

can you mess with Trp, Twr, Twtr, and Trtw settings?

I've tried 3-4-4-8 @ 260 with 3-2-2-2 which wouldn't even boot up (ran fine with TCCD).

i loosened them to 4-3-2-3 and it booted up fine.

@Painman: 250 3-3-3-7 won't work for me. The best I've seen anyone do is 3-3-4-8 or 3-4-3-8 (one of the two) but they had some awesome sticks.

I have the asrock 939dual-sata2. The only things that are adjustable are "Read preamble" (don't know what that is) and "Async latency" (presumably Trc)

 

[burney]

i got the g.skill ddr 4000 but i my stupid ga.k8nsc mobo could not handle it...

 

[Shimmishim]

Originally posted by: OvErHeAtInG

Originally posted by: Shimmishim
what board are you trying this out on?

can you mess with Trp, Twr, Twtr, and Trtw settings?

I've tried 3-4-4-8 @ 260 with 3-2-2-2 which wouldn't even boot up (ran fine with TCCD).

i loosened them to 4-3-2-3 and it booted up fine.

@Painman: 250 3-3-3-7 won't work for me. The best I've seen anyone do is 3-3-4-8 or 3-4-3-8 (one of the two) but they had some awesome sticks.

I have the asrock 939dual-sata2. The only things that are adjustable are "Read preamble" (don't know what that is) and "Async latency" (presumably Trc)

read preamble and asynch latency are different things.

have you tried A64 tweaker?

 

[OvErHeAtInG]

have you tried A64 tweaker?

No I'll give that a try.

 

[5t3v0]

Originally posted by: OvErHeAtInG
I've been priming since last night - Large FFT's on core 1 and blend on core 2. The blend errored out after exactly 8 hours 30 min - that's "stable" for me.

That's risky. Just because an error happens after 8 hours rather than 8 minutes doesnt mean its any more stable IMO. An error is an error & could happen at anytime & maybe when you need it least. I wouldn't be comfortable living with the knowledge that my OC wasnt 100% Prime95 stable.

In my test last night, Prime95 crashed after 2 hours at CL2.5 so back to the drawing board for me.

BTW, recommended setting for Async Latency = 7-8 & Read Preamble = 5ns according to a G.SKill technician. Higher numbers than this may give you more stability.

Watch out for A64Tweaker, I couldnt get it to work reliably at all. TREF is very hard to get right & the wrong setting can lead to BSOD & lockups.

 

[Duvie]

I am not so impressed about gskills doing 500ddr at cas 3....


I had some sticks that did 500ddr at cas 2.5...did 466 at cas2 and could hit near 600ddr with cas 3-4-4-8...

Thise sticks as well perfomred better with less volts....I bought them in Dec of 2004

I ultimately ran them at 520ish range cas 2.5,3,3,7 flawlessy with 2.7v (default) for 6-7 months

I guess I am more surprised they dont hit higher or hold 500ddr near cas 2 now.....I guess the technology really hasn't evolved that much....

 

[JSFLY]

Originally posted by: 5t3v0

Originally posted by: OvErHeAtInG
I've been priming since last night - Large FFT's on core 1 and blend on core 2. The blend errored out after exactly 8 hours 30 min - that's "stable" for me.

That's risky. Just because an error happens after 8 hours rather than 8 minutes doesnt mean its any more stable IMO. An error is an error & could happen at anytime & maybe when you need it least. I wouldn't be comfortable living with the knowledge that my OC wasnt 100% Prime95 stable.

In my test last night, Prime95 crashed after 2 hours at CL2.5 so back to the drawing board for me.

BTW, recommended setting for Async Latency = 7-8 & Read Preamble = 5ns according to a G.SKill technician. Higher numbers than this may give you more stability.

Watch out for A64Tweaker, I couldnt get it to work reliably at all. TREF is very hard to get right & the wrong setting can lead to BSOD & lockups.

Negative: If I consistently fail within 15 minutes of priming I will have a more unstable system than if I consistenly fail within 8 hours of priming. Read the Prime95 readmetxt.

Btw my Gskills go up to 260. I like em alot.

 

[nealh]

Duvie I agree to a point

you can not compare overclocks on the 512mg sticks to 1GB sticks..TCCD/TCC5 chips are amazing but can not be found on 1GB sticks
1GB sticks overclock less well

I got GSKILL still testing but it looks like 265 maybe my max in memtest86....still trying to optimize bios

I had hoped for 280..clearly I was expecting too much but these sticks need looser timing for overclocks

I have 2 sticks of 1 GB crucial valueram with 5G chips( not the famed binned ballistix) that do 250 mhz 3-3-3-7-1t with a venice 3200

my opty 170 will do 2600 1:1 260 with this ram but at 2700 I need to lower the ram speed to 225 as mem. controller is too stressed but I can run 2.5-3-3-8(I have not tried tighter yet with GSKILL)

High overclocks are not as common as I thought BUT the ram does as rated and is a good buy

also bios and ram slots is important..my SLI DR does not like ram in the orange slots(maybe a defective mobo) but loves the ram in yellwo slots

I had cold boot issues with orange slots not yellow and everyone recommends orange!

 

[superHARD]

OK, I guess I don't understand. You are running this memory at 3-4-4-8 at 2.6V all the way up to 260mhz...that's just 10mhz over stock mhz...I have never ever owned a stick of ram that would not OC an extra 10mhz...

 

[TrevorRC]

Duvie, I'm assuming you're talking about a 2x512MB kit with TCCD sticks in it [Those are the only kind of chips that can do what you're talking about...]

You can NOT compare 2x512MB kits to 2x1GB kits. End of story. The overclockability is COMPLTELY different.

The fact that a number of people [myself included] can hit DDR550 on these things with stock timings/voltage [and sometimes undervolted] is amazing, to say the LEAST.

Go buy a normal kit from OCZ or Corsair, let us know how high they can clock, and how much they cost

 

[nealh]

Originally posted by: superHARD
OK, I guess I don't understand. You are running this memory at 3-4-4-8 at 2.6V all the way up to 260mhz...that's just 10mhz over stock mhz...I have never ever owned a stick of ram that would not OC an extra 10mhz...

I agree about this..thus my disappointement but remember 1gb sticks are different

I had hoped 10-15% OC with no issue..or about 275-280

interestingly the newe chips ..060x GSKILL may(??) not OC as well as earlier batches

 

[5t3v0]

Originally posted by: JSFLY
Negative: If I consistently fail within 15 minutes of priming I will have a more unstable system than if I consistenly fail within 8 hours of priming. Read the Prime95 readmetxt.

Relatively, maybe so. However if you read the Prime95 stress.txt file, it proposes 2 schools of thought. I quote:

Most programs you run will not stress your computer enough to cause a wrong result or system crash. If you ignore the problem, then video games may stress your machine resulting in a system crash.... Keep in mind that the faster prime95 finds a hardware error the more likely it is that other programs will experience problems...

The second school of thought is, "Why run a stress test if you are going to ignore the results?" These people want a guaranteed 100% rock solid machine. Passing these stability tests gives them the ability to run CPU intensive programs with confidence.

If you are prepared to live with any error, regardless of how quickly it is found then there's a risk your computer will fail in real world use, especially if you do intensive stuff for long periods like encoding or gaming. At worst you could end with data corruption on your HDD. As with everything, ignoring an error is down to personal preference. My preference is not to.

 

[emilyek]

Let's not forget it is a 2 gig kit that will typically do DDR540 @ 7k+ mb/s for $160.

I have the Asrock and use a64 tweaker-- but just for the 183 memory divider. All the other settings (set to a GSKILL technician's specs) did nothing for my kit and board.

 

[nealh]

not alot are doing 270