DDR3 1333 OC to 1600 possible?

[Radio]

Hello, just a question on the above issue.

Im using a new build fo chess engine usage, which is more or less a full 100% load issue.

But, i am not sure im ram speed is making any difference there. Currently I'm using a 1300 pair which was available from another PC (no others available here).

Question is - do 1300s have the potential of a 1600 OC, just for testing the difference ? Or can no success be foreseen ?

Thanks.

 

[txus.palacios]

We need more info about those sticks. And about your computer.

 

[Radio]

Ok... its a 3930 K, currently oc to 4,4 Ghz, but i can go to standard for Ram-Comparison.
Corsair H110,
Asrockxtreme 3,
PS be quiet 480 W,
Standard Graphic, 512 Mb, theres neither OC nor any power necessary.
Win 7 Prof.

Unluckily i dont have the exact RAM description here, i could add that today evening.
Does that help, or does your answer especially depend from the rams ( i am afraid, yes...)

 

[txus.palacios]

Alas, we need info regarding your sticks.

 

[Radio]

Alas, we need info regarding your sticks.

Yup. That is, so far, already half of a good answer for me (some may work, some do not). It could have been worse.....

So, i will supply the Ram data in the evening and hope....

 

[Radio]

Ah, found them.

its TED 34096M1333HC9 , timing 9-9-9-24 @1,5V

Hope that helps. Thanks

 

[txus.palacios]

Those are Team Elites. I can't find specific IC info but they look like Elpida. Try 1.65v 1600 stock timings.

Also, I am not a very good overclocker, nor do I really tweak timings as I am not a bencher. We have a couple of friendly local memory gurus here, might want to wait for them to show up.

 

[Radio]

Those are Team Elites. I can't find specific IC info but they look like Elpida. Try 1.65v 1600 stock timings.

Also, I am not a very good overclocker, nor do I really tweak timings as I am not a bencher. We have a couple of friendly local memory gurus here, might want to wait for them to show up.

Thanks for your answer. I am OC newbie as well - would you mind to give some example for "1600 Stock timings " ?

 

[EarthDog]

Your stock timings are 9-9-9-24. He said to set the voltage to 1.65v with those timings.

 

[Radio]

Your stock timings are 9-9-9-24. He said to set the voltage to 1.65v with those timings.

Ups, now i get it, thank you so much !

Ill give a note how it works.

 

[txus.palacios]

BTW, that won't be a great difference, if I were you I'd just leave them running stock.

 

[EarthDog]

BTW, that won't be a great difference, if I were you I'd just leave them running stock.

This.

Performance increases are the biggest at that area, but are not really noticeable. Anything above 1600MHz returns even less.

 

[Radio]

This.

Performance increases are the biggest at that area, but are not really noticeable. Anything above 1600MHz returns even less.

BTW, that won't be a great difference, if I were you I'd just leave them running stock.

Yes, that is exactly what i need to test. The task is Chess Engine - which is VERY high load including parts of RAM. To be precise, the load is close to prime95 - levels. But i have no clue if most of the work happpens in L1/l2, or if standard RAM is involved.

I can say there ARE hashtables sized 256+ Mbyte involved, but i dont know how often they are consulted.

I can as well say that i see - seldomly - people where the search depths result just do not match the hardware (including oc) of the average, which are visibly very close to each other.

So it can be either an engine hack (mdifying engine behaviour) or some hardware issue i didnt understand yet.

One guy claimed it is because of his ram (2666).

And so you see me test it here - i dont want to permanently oc these TEDs, i only want to see if this makes adifference or not. If this difference (1333 to 1600) is not noticeable, then it is not the ram (because the observed differences are HUGH)

Stay tuned.

 

[EarthDog]

Just b/c a CPU runs at 100% doesn't mean it is memory bandwidth starved.

Will stay tuned!

 

[Radio]

Just b/c a CPU runs at 100% doesn't mean it is memory bandwidth starved.

Thats exactly what i have to figure out - if it is the case for me or not.

 

[Radio]

So, strangely the rams where running at 1,585 V, shouldn't that be 1.500 V ??...

However, with this Voltage they did run at 1600 without any noticeable problems. I have not made any real reliablility test, but a few minutes prime95 worked without any problems.

Unluckily, the real test i could do so didnt show up really reliable - the variation of the chess engine speedtest results have had too much variation to draw significant conclusion.
ill have to consider something more realiable (regarding the variation).
Thanks so far.

 

[EarthDog]

Did you use the blend test in P95? That is the one that tests ram...

 

[Radio]

Did you use the blend test in P95? That is the one that tests ram...

Yes, it was the blend test i used. But to be honest, i did not let it run long enough to get a rock-proof result, due to lack of time.

But i am not finished about this issue, there will be more ill check in detail and report here, if anybody is interested.

PS: i got the chess engine speed test reliable by enforcing hash clear for every test run. Normally the engine should do that theself.

 

[cullam3n]

I have my G-Skill 1333s running at 1600. Turns out they run just fine at 10-10-10-28.

 

[redduc900]

Have you checked the difference in bandwidth between DDR3-1333 at 9-9-9-24 and DDR3-1600 at 10-10-10-28?