[MemoryReaction] G.Skill F2-8000CL5D-4GBPQ - Low in cost, high in speed!

[EnJoY]

where do you get these for $114?

Thanks

The link is in my review, but here: http://www.newegg.com/Product/Produc...8000CL5D-4GBPQ

And for future reference, www.pricegrabber.com

[tomb18]

First thing you always MUST be doing is memtest86+. Current version is 2.0
http://www.memtest.org/

First pass tests 1-5, if no errors, try test 5 alone 10+ times. To iron out, pass the whole default set of tests 10+ times.

After this, try Linpack 64bit, problem size 30000, 100+ times. Good chance it will find some errors Memtest86+ missed.

P.S.
Don't forget to check the real vDIMM voltage. Many boards overvolt by default. Many ASUS boards overvolt by 0.05V.
And i wish you luck with 680i board, 680i is probably the most troublesome chipset for Intel CPUs.

Well it passes all of the memtests several times (but not yet the whole set 10+ times) but the machine locks hard within a couple of minutes with linpack.
This was not the case with my previous 800 MHz 4G ram.

Still trying some things....

[Cronos]

I hope you really need that SLI thing. Otherwise, get rid of this buggy piece of **** and get yourself decent P35/X38 board.

[Reinvented]

Well it passes all of the memtests several times (but not yet the whole set 10+ times) but the machine locks hard within a couple of minutes with linpack.
This was not the case with my previous 800 MHz 4G ram.

Still trying some things....

What hardware are you using?

And if you are going to run the memory at spec, don't run it Linked, always use Unlinked.

2.1v should be good for 1100 at least.

Also, did you do any mods to the board? like the vdrop mod? And what are your other settings.

[tomb18]

I hope you really need that SLI thing. Otherwise, get rid of this buggy piece of **** and get yourself decent P35/X38 board.

Well,
I've actually had really good luck with this system. Never had a problem, and I run my QX6700 at 3.5GHz.
I bought cheap ram in the past OCZ 6400 Platinum and they ran fine in 4G configuration.

As to linpack,
Isuspect that something more that just ram is stability is going on. Please seem to be having good success with a lot of the prime95's, Orthos etc but linpack just kills the systems.
IN my case I suspect it's something to do with my CPU overclock. I have never had a crash with any of the usual tools mentioned above, and in the year I have been using my system also.
What I find suspect is that my CPU temperatures never go above 60C with all of the above tests. I have a watercooled system with a triple rad just for the CPU. With Linpack, the termperatures hit 72C almost immediately. I think that I am hitting a CPU limitation and not ram although it seemed to run fine with my 4G set. Any insight?

[Cronos]

tomb18
Try lower your CPU multi. 72C is normal for Linpack 64b under water, i have the same temps and i can run Linpack without any lockups (on P35). But with errors, if memory is set to 1000MHz.

680i is very bad for multitasking and has very low memory performance in this case. It has never being designed for such a high load as Linpack.

[tomb18]

Well, I ran memtest overnight and this morning there was an error on test 5.
This is at stock memory settings 1000MHz unlinked, 5,5,5,15

The system specs are as follows:
QX6700 10x multi 1400 FSB voltage under load 1.45V
Water cooled (triple rad)
nvidia 680i board
4x2G F2-8000CL5D-4GBPQ 2.1V 5,5,5,15
Silverstone 1000W PS
All chipsets set at auto voltage

So I am now in the process of running the memory at 900MHz with memtest.

In Vista64, it seems that I can use the computer for everything including processing really large photos in photoshop (4G). If I run prime95 on 4 cores for a while the system locks hard. Can't move the mouse nothing. The same happens with linpack but much quicker.
This is something I never saw previously with my system. Normally, there would be an error in prime95 for example and that's it.

All right. It is possible that one or more dimms are defective given the failure on memtest. I suppose that if I can do 2 dimms at a time and see if there is a pattern perhaps there is a dimm that is not so good.

I do not really mind running at 800MHz, since the presence of 8G for my photoshop work is what really counts. I did some processing last night with a 48x72" 300dpi file with several layers and filters and it took about 10 seconds vs 45 minutes or so in the past with 4G.

A question: If this memory is failing at 1000MHz, is it a dimm fault or just an issue with speed with 4 dimms installed?
Thanks for your help

[Crazy Chuckster]

Well its a real toss up between buying these and the gskill pc 8500.......

[squito]

While these sticks won't do 500 on our Commando/E8400 setup due to that tRFC limitation, 475 is a major improvement over what we managed with a 4x1GB setup (Ballistix PC2 6400) ...

http://img136.imageshack.us/my.php?i...5gskillnl0.jpg

P.S. Memset always shows tRTP lower (-2) for some reason ...

[Cronos]

A question: If this memory is failing at 1000MHz, is it a dimm fault or just an issue with speed with 4 dimms installed?

Both i guess. The memory has too little OC headroom.

You see, the OC result seems to be inversely proportional to how well you test.
Windows/SPI =high OC.
Memtest86+ quick =average OC.
Memtest86+ long =low OC.
Linpack = lucky if stock.

Still suffice for many gamers i guess, just not for serious use...

[tomb18]

Ok I've done a little more testing.
First of all I loaded the latest Bios for my board (it was 3 versionsbehind)
I then ran memtest at 1000MHz 5,5,5,15 2.1V on test 5
and I am always getting an error at one location.
So I tested each module individually, and all are fine. Next I tested them in pairs and all is fine. Put all 4 back, and the failure occurs at the exact same address at test 99%
. I will now try at 800 MHz.

At 800 MHz, same error!
Lowered the FSB to stock, same error.
Upped the MCH voltage twice, no difference.
Now, I'm putting the voltage down on the memory to 2.0V.......fails
What's left?????
Now everything is at stock defaUlt. It still fails at the same place.
Last thing to try, is setting 5-5-5-12 at 800 MHz

[Cronos]

Test 5 cannot show the exact error location by its nature.

[tomb18]

Test 5 cannot show the exact error location by its nature.

It does, and every single time.
Address # 00076086a98. The data error changes though. And always at Pass 24%, Test 99%
I now have exhausted all possibilities.
I guess these are going up for sale. They work fine in pairs just not 8G

What a waste of money.

[Cronos]

It does, and every single time.
Address # 00076086a98. The data error changes though. And always at Pass 24%, Test 99%
I now have exhausted all possibilities.
I guess these are going up for sale. They work fine in pairs just not 8G

What a waste of money.

What if you change DIMM order, will error location also change?

If possible, try to test on P35/X38 boards.

[tomb18]

What if you change DIMM order, will error location also change?

If possible, try to test on P35/X38 boards.

Yes, the error will change. It is either one address or another. Only two addresses.

It almost sounds like a design flaw on the motherboard which doesn't show up with 4x1G ram.
But there are other people that are running 8G on the 680i without problems or so they say).

I am reluctant to get other ram, since I may run into exactly the same problem if it is a motherboard issue.

Man, I never had these issues with my dual opterons in the past.

[Cronos]

But there are other people that are running 8G on the 680i without problems or so they say).

This is the key What i have discovered, there are always ppl. who have great overclocks and i never could reach the same level. Why so, i asked myself?
With time, i just learned that great overckocks are always taken by the cost of some stability loss.

Can you live with some potential instability? Most ppl. can and thats what they are doing. Because most users do not load they computers that high or do not do any critical work. What's the loss if some pixels in your favored game are rendered in slightly wrong color? You will never notice this.

Don't expect desktop stuff, and especially overclocking memory, to undergo the same rigorous testing as server stuff.

[tomb18]

So Cronos, what do you think the issue is in my case. A defective address line or something glitchy in the traces or mch?

[Cronos]

No one will tell you.
I had 680i ASUS board with 4x 2GB Supertalent memory, based on Micron D9HNL, running DDR2-667@800 no problem (except very low memory performance, thats why i switched to P35).

You have two options : try to test on P35/X38 board. If you don't have/need SLI, i highly recommend you make this switch anyway.
Or try another pair of sets.

By the way, make sure your tRFC is right, should be 52+ for DDR2-1000 speed.

[tomb18]

Well I do have 2 8800GTX's in place for occasional gaming.
I'm thinking of getting the 780i board which apparently does support 8G of this memory as per G. Skill's testing.

[Ajarn]

Haven't stability tested yet (will get to that soon), but this is my max speed with a kit of these sticks so far at 2.1v.

[Cronos]

Haven't stability tested yet (will get to that soon), but this is my max speed with a kit of these sticks so far at 2.1v.

Who cares about max you can post? This kind of spam is really annoying.

Max stable speed is all that matters.

[EnJoY]

Who cares about max you can post? This kind of spam is really annoying.

Max stable speed is all that matters.

Not at Xtreme Systems. THIS kind of post really annoys me.

[Getttosmurf]

Not at Xtreme Systems. THIS kind of post really annoys me.

agreed this is not extremlystable systems :P

[Jodiuh]

Shoot me down, but I'll take my stability over 100mhz anytime.

It's fun watching you guys burn up parts in quest of crazy screenies tho.

[Cronos]

It's fun watching you guys burn up parts in quest of crazy screenies tho.

This may turn to a big disaster then someone, after reading too optimistic review, rush out and buy the product, just to find out it is not as good as he expected.

There is nothing fun about this.