G.SKILL Ripjaws 8GB F3-12800CL7D-8GBRH 1600 7-8-7-24 1.6 (2 x 4GB)

[Afrokalle]

Did some more testing today...

Super PI 1M 2000MHz 8-9-8-24 2T @ 1.65v:



Super PI 8M 2000MHz 8-9-8-27 2T @ 1.65v:



this memory is really impressive cause my i5 670 IMC usually have problem with 2000MHz

Tomorrow will switch to Gulftown and try to push a little bit more....

[youngpro]

some bandwidth comparison



This is based on maximum stable results at different primary timings

19885 MB/s read @ 2100 MHz 9-10-9-24 2T, 1.5v
18362 MB/s read @ 1900 MHz 8-9-8-24 2T, 1.5v
16437 MB/s read @ 1700 MHz 8-9-8-24 1T, 1.5v

it makes sense with a 2 or 3x 4GB kit to go with high TRCD 9-10-9 and get the high MHz for nearly 20K everest read MB/s, that is more than enough for any gaming system or power workstation

[youngpro]

very interesting results here...

I have added some Clarkdale vs Gulftown.. for a couple of game benchmarks Crysis and Heaven...

The Gulftown is the high end 980X with some high end Hyper Elpida modules clocked at 2000 MHz 7-7-7-20, with a Gigabyte 6850..

The Clarkdale is a very cheap part, on Gigabytes H55N-USB3 board with the ripjaws clocked at 950 8-9-8-24 with a Gigabyte 6850..

So there is no competition here right, the over $1000 980X will wipe the floor?

My clarkdale "gaming" system is
G.Skill Ripjaws 2x4GB 1600 7-8-7-24 clocked at 1900 8-9-8-24
Gigabyte H55N-USB3
Clarkdale @ 4GHz
Gigabyte 6850 @ stock
Windows 7 64Bit

Note: There is 20~MHz difference in the platforms that would slightly affect results in favour of the clarkdale, the gulftown also has 100MHz higher in ram clocks.





Clarkdale Heaven


Clarkdale Crysis


Gulftown Heaven


Gulftown Crysis


there is NOTHING in these results... i think resutls speak for themselves.. for gaming platform think cheaper with p55, clarkdale and some 2x2 or 2x4gb kits or AMD 890fx... for synthetic benchmarks, think x58

[Leeghoofd]

I would still love to have 4 real cores iso hyperthreaded ones James, especially in Battlefield Bad Company the difference is so noticeable.

Some questions :
Isn't the minimum FPS higher on I7 ?
What happens if you game at 1600 x1200 ? or lower ?
Maybe Crysis was totally GPU limited due to the chosen res ?

[youngpro]

yes its definatly totally GPU limited, cpu frequency doesnt make an impact here... but bro even with 3 5870 its the same LOL

Min and Max FPS on this setup were higher on Clarkdale, not really sure about what that is, but the difference is negligable, within 1 FPS which would be within a tolerance of acceptable change from run to run

I will rerun the Crysis benches are lower resolution.. is there a battlefield bad company benchmark?

[Leeghoofd]

One of the reasons I never liked Crysis as a standard. But since the general question pops up : Can it run Crysis ? I include it in my next GPU reviews.

No idea bout the battlefield one, but I can feel the difference while playing between my E8600 at 4.5Ghz and the Q9550 at stock... I'm getting some weird results too with Lost planet 2 (integrated benchie) Maybe talk to Mac, , think they got a demo or sortlike at Canucks...

[youngpro]

i think some of these games are more than likely very strange like the benchmarks we play with... maybe some prefer all cores and no threads, only scale to a specific number of threads (crysis?) and some are more gpu or cpu limited... i guess as someone that doesnt play games, i have picked just a couple to include in my analysis, but yes the "crysis question" always comes up.. if there is some better game benchmark for me to include, i can do that in the future also

[Leeghoofd]

Resident evil and Far cry 2 are pretty consistent James in output. I tried the latest Codemasters F1 1010 ( with latest patch ) but the game loads way too many different demos...

[youngpro]

RE5 right? you remember gigabyte beat me if you dare comp? haha i will stay away from RE5.. but Far Cry 2 yeah thanks bro I might add that to my list when I do future review, it is nice to add some benchmark that people other than overclocker can relate too

[PcCI2iminal]

Ive finished my review

http://www.criminalcafe.com/content....se-e-overclock

its portuguese but lots of screens

very nice kit
2100MHz LinX stable

http://www.youtube.com/watch?v=MX4kC...embedded#at=80

[Grinder]

I'm looking for 2X4GB for a thuban based daily use system for high end gaming/occasional benching. Are these a good bet? Anything better performing that I should consider in this price range? I know there's faster and more expensive, but haven't been into this for a while. Need to be available in Canada, and I see these for ~$180 + tax.

[chew*]

Should be good for thuban seeing as they work well on deneb and i would expect a bit more headroom on thuban providing you choose the right board.

[Grinder]

Thanks chew*, I was hoping you'd comment but didn't want to pester you with a pm.

I wonder is this the same chips as the 2X2gb f3-12800cl7d-4gbrh sticks that I have. And if they will perform ok when bulldozer hits on an am3+ (need a crystal ball to answer that one, whether they would be fast enough to feed bd). Anyway, I can't see anything else available here in that price range that would be nice. Even dom gt 1600c7 is the same price for 2X2gb as for the 2X4GB Ripjaws, much less $300 for c6 if you can find it. I know I'm limited by my 790XT anyway, but it's been a great value, taken lots of abuse and keeps ticking.



Edit: FYI, I read an aphnetworks review of these that says they use micron manufactured Hynix H5TQ2G83BFR chips. I ordered a set for $174 CDN. I'd like faster, but they seem to offer a nice blend of speed, price, and capacity.

[Erklat]

Care to fix this? The result is a low as 0.2 frames better (ie negligible) in the favor of Clark and yet it's bar is three times bigger. 95% of readers only look the charts. Then again you could say it's their problem but you might as well accept this little bit of criticism

[youngpro]

meh eheh as you said, the data is there.. i just suck at doing charts!

[Grinder]

No joy with these so far

Rather than push this thread OT, I posted in the MA790XT and GSKILL areas.

[chew*]

No joy with these so far

Rather than push this thread OT, I posted in the MA790XT and GSKILL areas.

I think board choice is sort of critical here.

They work good in my ASUS 890 boards, yet to test in the gigabyte ud7 2.1 but i imagine they will do well there also.

790 boards well, i would expect them to not fare as well.

[Grinder]

Well, I only have the one board so I can't really tell if it's bad sticks or board limitation.

I expect you are right, but I was hoping I would achieve the spec of 1600 787 rather than the numbers you posted, since the board isn't optimal but I'm on Thuban.

I'll tinker for a bit and ask GSkill for help. If still no joy I'll rma or buy a chiv, but I was hoping to hold off till am3+.

I know, I know. You said "providing you choose the right board" in your original response.

[chew*]

Well, I only have the one board so I can't really tell if it's bad sticks or board limitation.

I expect you are right, but I was hoping I would achieve the spec of 1600 787 rather than the numbers you posted, since the board isn't optimal but I'm on Thuban.

I'll tinker for a bit and ask GSkill for help. If still no joy I'll rma or buy a chiv, but I was hoping to hold off till am3+.

I know, I know. You said "providing you choose the right board" in your original response.

you can try what i had to do on previous 790 boards with certain modules if shooting for 1600.

drop to 1333 divider, jack fsb up to 240. Let me know how that works out for ya.

[Grinder]

you can try what i had to do on previous 790 boards with certain modules if shooting for 1600.

drop to 1333 divider, jack fsb up to 240. Let me know how that works out for ya.

I was already using the 3/8 divider with ref of 308 for my fixed mult oc on 1055t (4004 cpu, 1642 tight ram)... I never did get the 1600 divider to work for me with the 2X2GB Ripjaws.

I'll try bumping IMC volts, and test sticks individually (as GSKILL suggested). I suppose if that doesn't work I can drop below 308, ++ cpu mult and look for a new sweet spot if I can find what makes the ram happy.

Update: I loosened the timings to the SPD values display in the BIOS, i.e. 1600 1.6v 999 160ns 2T et cetera, and bumped cpu nb vid to 1.35 @ 2700 (from 1.3) . This would pass one round of memtest86+, but bluescreen and reboot while loading windows. I've put my 2X2GB sticks back in for now since I have some work to do. I'll test individual sticks in 4 or 5 hours.

Also, my apologies to the OP, since these aren't exactly benchies showing off what these sticks can do...

[Grinder]

From my post in the GSkill section:

Invidual sticks will both boot windows if I set xmp timings, I think it was: 1600 9-9-9-24-34 160ns 2T 1.6v.

One stick tested with a quick round of memtest+, then boot windows and 3 hours of lite gaming (q3 mod) and browsing, as well as 5 min of prime blend.

The other stick I ran 7 hours of memtest, then booted windows and ran 5 min of prime blend. So both need more testing to be certain they are completely stable.

If I tightened up the sub timings on either stick, they would blue screen on windows load. I still haven't found what settings will work with both sticks installed. But the idea was to have 8GB of ram that was as fast or faster than my 4GB @1642 777 1T...

Does this likely indicate a limitation of the 790 chipset? I don't think it is my thuban IMC, since bumping volts didn't help, and I think I tried dropping the freq to 2000 (currently 2700, normally almost 2800 with the 4gb kit).



chew*, does it seem likely these would play nice in a chiv? i.e. this behavior is consistent with 790 limitations rather than an inferior kit? I know you can't say for certain, but I don't want to give in and buy one, then find they still won't work well.

[chew*]

chew*, does it seem likely these would play nice in a chiv? i.e. this behavior is consistent with 790 limitations rather than an inferior kit? I know you can't say for certain, but I don't want to give in and buy one, then find they still won't work well.

testing them now in 32m in C4E with 1090T

[Grinder]

testing them now in 32m in C4E with 1090T

Look forward to your results. I just broke down and ordered a C4F to pair with these and 1055T, will give that a shot before rma. May be a couple weeks before I have time to play, so I'll watch your results with interest

[chew*]

I did not run spec.....Keep in mind this is only 32m. Everything i have tested on amd as far as prime stability has not exceeded 2100 prime 95 stable.....there is however a wall for 32m at 2200 on AMD it seems so I pushed to see if it could max out to the wall.

Untuned vista, ran just to verify stability.

Prime stability will be tested in my full review.

[DGLee]

Wow. I'm really being curious about it's OC headroom...... are you aware of the module used for that 4GB x 2 kit?
RH = still Elpida's?