G.Skill F1-4000BIU2-2GBHV Infineon CE-5 OCing Report!

[bachus_anonym]

Yo XS Bandwidth Freaks!

Welcome to yet another episode of OCing Reports! Few days ago I got the package from G.Skill USA with their new 2GB set - F1-4000BIU2-2GBHV which has rated speed of DDR500 8-2-3-3.0-1T. This set is built with Infineon CE-5 DRAM components. By lifting heatspreader a tiny little bit I managed to see the markings to confirm that as well as IC weekcode of 0550. As usual, modules are built on Brainpower B6U808.

This write-up is intended to show what this new product may be capable of. I have seen few other members posting results with this set in XB section, including CapFTP G.S EUTech and quent. So, you will have some basic idea about the product already, even though all sets were provided directly by G.Skill. Initial and very small shipment of those have been sold out at Newegg but I was told there's more on the way from G.Skill USA warehouse.

I will be testing stability with SuperPi 32M, Memtest for Windows, SP2004. As we already know, 3D testing seems to be mandatory for Infineon-based modules, therefore I will also run 3DMark2003 and play NFS:MW as well as FarCry.

Setup:
Opteron 146
DFI NF4 Ultra-D, BIOS release 11/14/2005
EVGA 7800GTX 256MB KO @ 540/1400 ForceWare 81.98 & 84.21





Let's get going...

Part1: ALL-STABLE, including 3D benchmarking and 1hr of gaming

SPi32M & 3DMark2003 @ 300x10=3000Mhz DDR500 5-2-3-3.0-1T

SPi32M & 3DMark2003 @ 306x10=3060Mhz DDR510 5-2-3-3.0-1T

SPi32M & 3DMark2003 @ 309x10=3090Mhz DDR514 5-2-3-3.0-1T

2hrs of SP2004 LargeFFT @ 309x10=3090Mhz DDR514 5-2-3-3.0-1T

~600% of Memtest for Windows @ 309x10=3090Mhz DDR514 5-2-3-3.0-1T

SPi32M & 3DMark2003 @ 312x10=3120Mhz DDR520 5-2-3-3.0-1T

~600% of Memtest for Windows @ 312x10=3120Mhz DDR520 5-2-3-3.0-1T


The maximum fully-stable OC I managed to achieve seems to be 260MHz and is is pretty much the same as other revied kits did. I could reliably play both games without any halts, freezes or artifacting. As soon as I raised to 261MHz, lockups started with looping sound, etc. More voltage (up to 2.9v) did not ease the situation. Findling with latencies and other DRAM settings in BIOS did not help either. In other words, I got typical symptoms that I have experienced in the past with other Infineon modules (Corsair's PC4400PRO, OCZ's EB PC4000 and GSkill's PC3200ZX) where past certain speed (vary for different sets) system looses 3D stability.


I also run some test at CAS2.5. To my surprise, it was absolutely neccesary to bump voltage to 2.94v (2.98v with DMM) to achieve stability at DDR500. Without this, system refused to boot to Windows no matter what DRAM settings were used. But once there, I managed to run a full set of my stability tests:

SPi32M & 3DMark2003 @ 288x10=2880Mhz DDR480 5-2-3-2.5-1T

SPi32M & 3DMark2003 @ 300x10=3000Mhz DDR500 5-2-3-2.5-1T

2hrs of SP2004 LargeFFT @ 300x10=3000Mhz DDR500 5-2-3-2.5-1T

~600% of Memtest for Windows @ 300x10=3000Mhz DDR500 5-2-3-2.5-1T


A small update with CAS2.0 results. I spent about 30mins trying to get div150 working with CAS2.0 (no problem with 2.5 or 3.0) but for some reason it would just not boot to Windows. I will try again later but for now I decided to drop HTT a bit and go with div166.

SPi32M & 3DMark2003 @ 264x10=2640Mhz DDR440 5-2-3-2.0-1T

2hrs of SP2004 LargeFFT @ 264x10=2640Mhz DDR440 5-2-3-2.0-1T

~600% of Memtest for Windows @ 264x10=2640Mhz DDR440 5-2-3-2.0-1T


Part2: Desktop stable only

SPi32M @ 318x10=3180Mhz DDR530 5-2-3-3.0-1T

SPi32M @ 322x10=3220Mhz DDR536 5-2-3-3.0-1T

2hrs of SP2004 LargeFFT @ 322x10=3220Mhz DDR536 5-2-3-3.0-1T

~600% of Memtest for Windows @ 322x10=3220Mhz DDR536 5-2-3-3.0-1T


Part3: Ehhh... My conlcusion

Well, there you go! G.Skill finally released DDR500 rated Infineon based 2GB kit. After PC3200ZX and PC3200HX, it was about time to do some higher binning. I was hoping for a bit more headroom in 3D testing, but DDR520 is still very satisfying. I'm very interested to see how the rest of the 2GBHV supply is doing. So, go get one and report back here, OK?

I will try to answer any questions regarding above results. Fire at will!

Thanks for showing interest in this OCing Report

[Coroner Kyle]

Nice review as always.

What's the max voltage that CE-5 can handle without the chance of them dying? I see the sticks are rated at 2.6-2.8V at Newegg.

[ozzimark]

just for kicks, you mind giving 3-4-4 a go?

[Raybo]

Last thing I need to do is buy new sticks.

So your review is good news to me!

[bachus_anonym]

What's the max voltage that CE-5 can handle without the chance of them dying? I see the sticks are rated at 2.6-2.8V at Newegg.

Honestly, I have no idea! During SP2004 LargeFFT and M4W @ 2.98v (DMM) I was poking around with temp probe, touching heatspreaders, sticking it underneath and touching ICs. I have literally not seen temperature much higher than ambient. At 18C room temperatre, I saw 23-24C. So, if temperature is a dying factor, I would say they can withstand some voltage. But I have no idea if there is something else than can contribute to damage.
Most testing I could complete within 2.5-2.7v but I was shown some pretty insane voltages (BH-5 style!) on other Infineon CE-5 modules - not GSkill though. If they survive in a long run, I don't know.

just for kicks, you mind giving 3-4-4 a go?

Sure I don't expect 3D stability there (I have passed SPi32M with 2GB Corsair PC4400PRO CE-5 @ 287Mhz 8-3-4-3.0-1T) but I will see how far it will bench 32M at those latencies

[Raybo]

Last thing I need to do is buy new sticks.

So your review is good news to me!

I have to quote myself here.

OTOH, if I could tighten up my timings............

[ozzimark]

thanks bachus

though tbh, considering CE's nature, i'd rather 3d benches.. 3dmark is faster than 32m isn't it?

[bachus_anonym]

thanks bachus though tbh, considering CE's nature, i'd rather 3d benches.. 3

Then I can tell you even now, that 8-4-4-3.0-1T is not gonna increase 3D stabililty At 272Mhz, even 3DMark's splash screen has hard time showing up, not mentioning actually running it

[Coroner Kyle]

I think my Corsair 3500LL's are CE-5 and they do not get hot. I set to 3V just to see if they would get warm and nope, but as you said b_a I wonder what it is that could require the low volts with their ability to run low temps.

I'll try to post some of my results. Have you tried any tighter timings with lower speeds?

[ozzimark]

Then I can tell you even now, that 8-4-4-3.0-1T is not gonna increase 3D stabililty At 272Mhz, even 3DMark's splash screen has hard time showing up, not mentioning actually running it

all right, that's all i really needed to know, thanks

[high5]

great stuff bachus... as always!

This set is built with Infineon CE-5 DRAM components. By lifting heatspreader a tiny little bit I managed to see the markings to confirm that as well as IC weekcode of 0550.

so you're saying the ICs are actually not re-labeled like on HX?

So, if temperature is a dying factor, I would say they can withstand some voltage. But I have no idea if there is something else than can contribute to damage.

bad side effects of excessive voltage happen regardless of the operating temperature, but high temp accelerates those effects.

so bottom line, you need much time to see whether that Vdimm actually hurts them.

i'm interested in one more thing; out of all Infineons you played with, how much MHz have you actually gained (on avg) by further relaxing timings from 3-3-2?

[El Snorro]

Nice review Bachus!

I have tested this kit myself and I liked it also!

[stevke]

Nice review, I think I'm going to buy a pair

[kakaroto]

Very nice testreport bachus_anonym, as always.
I think all 2GB HV have average op 260-270 3-3-2-5 windows and 3D stable.

[bachus_anonym]

so you're saying the ICs are actually not re-labeled like on HX?

Yes, ICs have their original markings left unaltered, from what I could see.

i'm interested in one more thing; out of all Infineons you played with, how much MHz have you actually gained (on avg) by further relaxing timings from 3-3-2?

Nothing extra in terms of 3D stability, unfortunately... Also, past certain point (around 10MHz higher than 3D stable) I could notice slight corruption on a Screensaver Tab of Display Properties. A lot of dots on that "thumnail". When pushed really high, too high for even some desktop stability tests, occasional screen blinking could be noticed - it goes blank for a second and comes back.

I have tested this kit myself and I liked it also!

Yeah, it's not bad at all G.Skill fans will have something to look for in the stores.

I think all 2GB HV have average op 260-270 3-3-2-5 windows and 3D stable.

That's what it looks like by looking at other tested kits...

[high5]

so you'd agree that generally w/ Infineon's there's no need to relax Trcd & Trp beyond 3 & 2?

slight OT: dunno if you noticed, but TWINX2048-4400PRO is listed again on Corsair's website.

[bachus_anonym]

I'll try to post some of my results. Have you tried any tighter timings with lower speeds?

UPDATE: I have just added CAS2.0 results to my initial post

so you'd agree that generally w/ Infineon's there's no need to relax Trcd & Trp beyond 3 & 2?

From what I experienced, no If 3D stability is what one is shooting for. It seems that is not latencies that made Infineon unstable but pure memory frequency. I would really like to know why those Infineon parts behave like that in general. Oh well, like you said earlier - just another feature and something to get used to

slight OT: dunno if you noticed, but TWINX2048-4400PRO is listed again on Corsair's website.

Yep, they finally announced it with press release right before Cebit. I'm glad it finally sees the light, after almost 6 months since first previews

[CapFTP]

good bachus....nice to see we are testing the same....

I didn't work so much on voltages...so @ cas 2.5 I just hit 242 fully stable.

The only test I did with 2.9-3.0 v was to see if there's some improvment for max freq@ cas3 (270 stable also in 3d appl...but nothing changed..)

I'll try as you to push 250 cas 2.5 with more volt...

Very good result...and I appreciate the same structure of the 1st post....is the same idea I had to summarize every test..

[bachus_anonym]

I didn't work so much on voltages...so @ cas 2.5 I just hit 242 fully stable. The only test I did with 2.9-3.0 v was to see if there's some improvment for max freq@ cas3 (270 stable also in 3d appl...but nothing changed..) I'll try as you to push 250 cas 2.5 with more volt...

Very good result...and I appreciate the same structure of the 1st post....is the same idea I had to summarize every test..

I have not experienced any gains in 3D stability at CAS3.0 past 2.7v. Only when switched to CAS2.5, extra voltage seemed to help and 2.98v was what worked for DDR500 5-2-3-2.5-1T - both desktop and 3D. But I would think that 2.5-2.7v is the range in which this kit works best.

I like to make my OCing reports easy on the eyes so I put results into groups and use links instead of pasting screenshots. Just make it easier to go thru without eating bandwidth

[OBR]

This RAM is really worst than GSkill-HZ, my max OC with HV is 260@6-3-3-3, and 240@CL2.5, for that price is this memory VERY VERY BAD.

[GSKILL TECH]

This RAM is really worst than GSkill-HZ, my max OC with HV is 260@6-3-3-3, and 240@CL2.5, for that price is this memory VERY VERY BAD.

i would said yeah this ram seem like not big OC bur very good compatible. with most of 2GBHZ "2T" mobo 2GBHV can do 1T no problem
and it is the "REAL WORLD KIT". NO MORE HANDPICK SAMPLE is better than later on 95% of user can not do as high as review kit

[bachus_anonym]

This RAM is really worst than GSkill-HZ, my max OC with HV is 260@6-3-3-3, and 240@CL2.5, for that price is this memory VERY VERY BAD.

I remember you were dissapointed that your UCCC did not do DDR600... I think, your expectations are simply very high but reality is a bit brutal Although, we have seen higher clocks on CE-5/6, DDR520 on that set is not what I would call "very very bad". It's not exceptional but it's still good enough for people to find it interesting. I bet that someone out there will see numbers closer to 265 and maybe even more (in 3D) with this kit. Time and yields will tell

[diablos]

look at this

[stone_cold_Jimi]

In the absence of Xtreem Cronus, this looks like it would do. I got as far as filling in my details for some of this earlier and turned back, thinking that Ballistix and HZ kits were enough and I was Mr Greedy. But I'm not confident in the Ballistix; if it doesn't respond to a new fan, it's HV for me.

[Sam666]

Nice reports.