Xtreme Twinmos Memory!

Well unfortunately I got a bad pair of modules...

Wouldn't even run at the rated speed or timings, no matter how many volts, and just barely eke'd by at 200mhz 2.5-3-3-8 at 2.8vdimm...

Not really much point in even trying to burn them in, they'll just go back rma, and I'll have to try to find another pair or another type...

*sigh*

Incidentally it was ko-6633 pcb, and bh5 in the SP 3500 flavor...

BTW, the place I saw the BH5 SP on BP pcb was a seller on ebay, who'd put pics up of the chips and PCB, which confirmed it was brainpower, but the listing's gone and I can't find it unfortunately...

Ah well...we'll see how good Komplett are with the RMA procedure...

Gray

Quote:

---

Originally Posted by Gray Mole

--
BTW, the place I saw the BH5 SP on BP pcb was a seller on ebay, who'd put pics up of the chips and PCB, which confirmed it was brainpower, but the listing's gone and I can't find it unfortunately...
--

---

Have to say I'm a little sceptical about this sole appearance of TMSP with "BH-UTT" on BP.
The must have been some kind of confusion. / Scam attempt.

Quote:

---

Originally Posted by largon

@LANero:
Trfc: 12
Twr: 2
Trtw: 2 or 1
Tref: On CH-UTT this does alot, try different value.
Read Preamble: 6ns
Idle Cycle Limit: 0 - 16 clks

---

Hello largon

You recommend some value in special for Tref, lower perhaps?

The Idle Cycle Limit helps to improve the performance?

Thanks for everything

260 Mhz@3.4v - LTD 3x@1.30v - Vcore@1.525+110% = 2340Mhz on a +3000 still 1:1

Suggestion dont use vantec heat spreaders coz mine trapped the heat and it sucked, is there a way to improve these heat spreaders

@LANero
Maybe try 0310P BIOS and 1,5-2-2-6/7 @260 MHz. :slobber:

Quote:

---

Originally Posted by BitpowerPM

@LANero
Maybe try 0310P BIOS and 1,5-2-2-6/7 @260 MHz. :slobber:

---

BitpowerPM.

You think that you am better option 3/10P?

Single I have tried it once and he gave the impression me to go better this.


I would have to go back by 3/10P?


Thanks

I just won a pair from ebay in the uk the same as yours.
ko-6633 pcb and i think week 19 year 2005 in cpu-z

I have them runnig at 210fsb 2.8v 2-2-2-7 5:4 divider in a ic7 max3 with vtt and vdim mod. I havent tried any higher vdim yet until i get better cooling for them.

I also have some corsair xms 3500 v1.1 with bh-5 but then are 256mb sticks I had them upto 255fsb 3.1v 2-3-3-7 in the same board.

Quote:

---

Originally Posted by Gray Mole

Well unfortunately I got a bad pair of modules...

Wouldn't even run at the rated speed or timings, no matter how many volts, and just barely eke'd by at 200mhz 2.5-3-3-8 at 2.8vdimm...

Not really much point in even trying to burn them in, they'll just go back rma, and I'll have to try to find another pair or another type...

*sigh*

Incidentally it was ko-6633 pcb, and bh5 in the SP 3500 flavor...

BTW, the place I saw the BH5 SP on BP pcb was a seller on ebay, who'd put pics up of the chips and PCB, which confirmed it was brainpower, but the listing's gone and I can't find it unfortunately...

Ah well...we'll see how good Komplett are with the RMA procedure...

Gray

---

Even after weeks of burn-in, my TMSP 3200 (K0-6633) sucked pretty bad also. In SC mode, one module would do 233FSB 2/2/2/6 @ 3.4V while the other would do 223FSB 2/2/2/6 @ 3.5V. In DC mode, they could not do much more than 200FSB @ 2/2/2/6 @ 3.2V. At higher FSB and higher voltage in DC, they would be stable for 3-5 passes in memtest86 and then produce hundreds/thousands of errors. Overall, they could barely manage 200FSB 2,3,3,10 @ 2.8V. It would become stable when voltage >= 2.9V. Very marginal set of memory. Oh well, luck of the draw.

However, my two sets of Mushkin Blue are starting to respond finally. I was starting to despair because I thought I had purchased 3GBs of bad ram.

Cjoe

I couldn't even run the 2.5-3-3-8 at 200htt with them together...

Awful, really, I was expecting they'd at least run the numbers, and maybe burn in, but if I have to burn them in just to get the rated speed, there's something wrong...

I requested that they replace them with the SP CH5 utt's instead, and I think they're happy to do that...

Yeah, 1 bad GB is bad enough, good to hear the blue's are coming together for ya :)

Gray

I think many errors are user errors or bad hardware.
I have tested 20 BH-5 last week single channel and worst one did 258fsb superpi 1 meg.
Not one bad stick.

Quote:

---

Originally Posted by Ubermann

I think many errors are user errors or bad hardware.
I have tested 20 BH-5 last week single channel and worst one did 258fsb superpi 1 meg.
Not one bad stick.

---

Uh huh...

I've got 2 other sets of ram I like...

2*512 CH5 ocz and it runs 245htt at 2-2-2-5

2*256 BH5 Twinmos and it runs 255 at 1.5-2-2-5 or 260 at 2-2-2-5

User error? :slap:

Bad hardware? Oh yeah, it just happens to go faulty every time I throw these particular sticks in :rolleyes:

Gray

Quote:

---

Originally Posted by Ubermann

I think many errors are user errors or bad hardware.
I have tested 20 BH-5 last week single channel and worst one did 258fsb superpi 1 meg.
Not one bad stick.

---

You tested a lot of sticks. Did you do burn-in on any of those sticks? Sounds like you got a good batch.

To a certain degree, you maybe right because there are a lot of noobs out there. However, in my case and Gray Mole's, I will disagree with you. We just happened to get stuck with inferior memory.

I have two CPUs (3200+ and 3000+ topless) and two PSUs (Antec NeoPower and Fortron BlueStorm 500). I have tried various permutations on my DFI NF4 with all my memories (2x512 GSKill LE @ 300FSB 2.5/3/3/7, 4x512 Corsair XMS XL 240FSB 2/2/2/5, 4x512 Mushkin Blue (Green) @ 250FSB 2/2/2/6, 2x512 Twinmos SP default) at 1T 1:1. The TMSP would just barely work at default setting (an error would showup after every 20-30 passes in memtest). Even after significant loosening of the timings, it still wouldn't work very well in DC mode.

I even tried burning them in on my other rigs (DFI NF2 LP Ultra B, Abit IC7-Max3, Shuttle SN45G, and ECS 875 system) at default settings. They just weren't happy.

CJoe

Dont take it personal. Seen so many complaining about ohh this stick sux im going to RMA it.
And after some help and trying they usualy get it running pretty good.

Quote:

---

Originally Posted by Ubermann

Dont take it personal. Seen so many complaining about ohh this stick sux im going to RMA it.
And after some help and trying they usualy get it running pretty good.

---

I didn't take it personally. I was just trying to describe my testing approach and scenario. I read through all the threads and tried all the tips especially Largon's. Some worked and some didn't work.

My AA4T CH5 K0-6633 would not do > 210FSB @ 3.2V DC mode with tight timings of 2/2/5/2/7/15/2/2/1/1/1168/1 1T 1:1. It would do slightly better with looser timings but it was still unstable at any voltage.

BTW: How did you manage to burn in so many sticks in one week? Did they all get to 250FSB+ really easy? Most of my memories required about 3-4 days per stick to achieve total stability with no errors in Memtest, OCCT, and S&M.

Cjoe

Hey guys anything that I could improve here

Quote:

---

Originally Posted by cjoe

I didn't take it personally. I was just trying to describe my testing approach and scenario. I read through all the threads and tried all the tips especially Largon's. Some worked and some didn't work.

My AA4T CH5 K0-6633 would not do > 210FSB @ 3.2V DC mode with tight timings of 2/2/5/2/7/15/2/2/1/1/1168/1 1T 1:1. It would do slightly better with looser timings but it was still unstable at any voltage.

BTW: How did you manage to burn in so many sticks in one week? Did they all get to 250FSB+ really easy? Most of my memories required about 3-4 days per stick to achieve total stability with no errors in Memtest, OCCT, and S&M.

Cjoe

---

I never burned them, just tested them in superpi and sold them.

Doing some further testing on my new 2x512MB TwinMos SP PC3200. Looking pretty good @ 3.2v on my socket 754 setup. Doing 250mhz with 2-2-2-8 timings which brings me joy. They seem to perform about the same as my PQI TCCD's at about 270mhz with 2.5-3-3-8 timings and the same cpu clockspeed. I was thinking that the SP would hold a little bigger performance advantage, but I imagine it could be some tweaking I just need to do.

http://home.comcast.net/~jhatfie/superpi_1M_2750.jpg

Quote:

---

Originally Posted by Ubermann

I never burned them, just tested them in superpi and sold them.

---

Okay. Now I understand. Are you equating SuperPI @ 1M testing with total system stability at that FSB?

I can easily run a suicide SuperPI @ 1M at higher FSB in DC but I know the machine is not memtest, OCCT and S&M error free stable.

cjoe

No not stable but the few sticks i tried memtest on was stable ~ -10fsb of that superpi.

Quote:

---

Originally Posted by Ubermann

No not stable but the few sticks i tried memtest on was stable ~ -10fsb of that superpi.

---

Okay. That sounds more reasonable. How long did you test for? Memtest frequently runs clean for upto 10 passes and then generates errors.

Also, you're not encountering DC mode and 1T issues with you're NF3 board. Two single sticks doing 245FSB reliably probably won't do 245FSB in DC and 1T with the same voltage and same tight timings. I have 4 sticks of Mushkin Blue CH5 that will do > 245FSB with tight timings and > 253FSB with slightly looser timings. However, if I take the best two sticks, I can only achieve 235FSB with tight timings + 3.4V or 250FSB with looser timings + 3.6V.

You may have gotten a great batch of memory. :toast:

Maybe both my CPUs have weak memory controllers. Maybe I have two crappy power supplies and maybe my motherboard is not so reliable. Maybe, like you said "user error". (I could be doing something wrong). Finally, I may have crappy luck. :confused: Ultimately, you get what you pay for.

Fortunately, my GSkills do great with my 3000+ Winnie; otherwise, I would be seriously depressed. I should never have purchased those 3GBs of cheap memory because the odds were against them achieving > 265FSB. Any speed less than 260FSB is not significant. I should have purchased the OCZ VX which would have guaranteed ~ 265FSB. Its cheaper than the price of 3GBs of value memories.

Cjoe

If you get errors after 10 pass it is most likely heat problem i think.

Quote:

---

Originally Posted by Ubermann

If you get errors after 10 pass it is most likely heat problem i think.

---

I leave the case door off during burn-in and have a 90mm fan blowing directly on the memories. So, I don't think its a heating issue. Its not consistent. I notice that memtest will run clean for a long time with no errors. Then a few errors will crop up and it will be clean again for awhile. Then a couple of errors will pop up again. I've seen this happen on several of my machines as well as others.

I am looking at two different machines right now: DFI NF4 - 43 Passes with no errors and DFI NF2 - 21 Passes with 2 errors during pass #5. Its clean upto pass #5 and clean after pass #5. Its the normal random glitch as the memories get stressed.

My general testing methodolgy is to run a minimum of 10-20 clean passes with memtest before increasing the FSB. After the memory test clean, I usually verify by booting into windows and running OCCT and S&M. If both complete successfully, then I'm golden at that FSB.

Cjoe

ifu get some thart works at 5 volts send to me.quit playing with cheap ram and buy some thats guranteeed to run fast.I c the posts here 2-3-2-2.I run thios ram at those timings wannna c prooof and it still suks.oppps sry fer suyks but it does

...

shrooom that is direst to u big boy u trade me ur the one bragiing in both newegg and here gimme ur ram



------------
Edit by: bias_hjorth
I let this one stand. Please calm youself.