TwinMOS Speed Premium BH-die Review

[g0dM@n]

Here is my Refresh Rate through Sandra. As you can see it says 5.00uS, not 5ns like you would expect. That is why I just posted the pic b/c now I'm confused. I thought you need to know uS value... I didn't know that 5.0 was a real choice... I thought that it was supposed to be 3.9, 7.8, 15.6....

I guess it's supposed to mean 5ns, so 7.8uS?

 

[el]

yeah mine says 10us and I have the same ram. Any other way to see the refresh rate?

Below is basically my shot at 2.8ghz. It is game okay but not 24/7 stable adn that is what I really want. Hopefully Hellfire will release his great new bios and I will be very happy. I am learning tons but 50% overclocks are never easy.

==========


Genie BIOS Settings:

FSB Bus Frequency - 311
LDT/FSB Frequency Ration - 3
CPU/FSB Frequency Ratio - 9
PCI eXpress Frequency - 100Mhz

CPU VID StartUp Value - 1.425

CPU VID Control - 1.300v
CPU VID Special Control - Avove VID * 123%
LDT Voltage Control - 1.40v
Chip Set Voltage Control - 1.70v
DRAM Voltage Control - 3.20v

DRAM Configuration Settings:

DRAM Frequency Set - 140 (DRAM/FSB:7/10)
Command Per Clock (CPC) - Enable
CAS Latency Control (Tcl) - 2.0
RAS# to CAS# delay (Trcd) - 02 Bus Clocks
Min RAS# active time (Tras) - 05 Bus Clocks
Row precharge time (Trp) - 02 Bus Clocks
Row Cycle time (Trc) - 07 Bus Clocks
Row refresh cyc time (Trfc) - 13 Bus Clocks was at 14
Row to Row delay (Trrd) - 05 Bus Clocks
Write recovery time (Twr) - 02 Bus Clocks was at 3
Write to Read delay (Twtr) - 02 Bus Clocks
Read to Write delay (Trwt) - 02 Bus Clocks was at 5
Refresh Period (Tref) - 1816 Cycles
Write CAS Latency (Twcl) - 1 was at 5
DRAM Bank Interleave - Enabled



DQS Skew Control - INCREASE
DQS Skew Value - 0
DRAM Drive Strength - Level 6
DRAM Data Drive Strength - Level 3
Max Async Latency - 7 nano
Read Preamble Time - 5nano
IdleCycle Limit - 256 Cycles
Dynamic Counter - Disable
R/W Queue Bypass - 16 x
Bypass Max - 07 x
32 Byte Granularity - Disable(8 Bursts)



==========

 

[felinusz]

I'm also using Hellfire's 510 BIOS. I can't really tell the difference between it and Oskar's 510-2, but it doesn't seem any worse.

Have you tryed changing: tRFC to 13 and 16, tWR to 2, tRWT to 2, tWCL to 1. Those are all tighter than what you've got, but your UTT should handle them, and will perhaps see a small stability gain. In particular, a tWCL of 1 (or AUTO, same thing) seems to be important for my sticks (BH-5 and UTT are different, but you will hopefully find that it works).

 

[g0dM@n]

El, dude, try my timings... why are you running your ram at only 219mhz?

All of my settings are in my sig, and I'm on 510-2. Wait, maybe you don't have active cooling. If that's the case then I dunno if you can use a higher divider, but DEFINITELY your ram can do 240+... you just may need 3.3v to hit 240, though.

BTW, does anyone know a better way to find out the refresh rate? 5uS doesn't sound right... "uS" goes by multiples of 3.9 I thought, and "ns" goes by whole numbers...

 

[el]

gotta love this stuff! tighter equals more stable. yeah I am going to give these sticks more juice 3.3 or 3.4 and see what that gets me. I have the weird timing/divider cuz I was having major cold boot issues from day one with three different bioses and I finally settled on that weirdness giving me the greatest shot at 2.8ghz dvd shrink stable/and gaming of course.

Basically the tref and other advice really helped me get super stable but I run memtest and see. also wouldn't mind running 1.5-2-2-5@240 but I think 3.4volts will be needed.

 

[g0dM@n]

I don't trust the refresh rate value that Sisoft Sandra gave me. It just doesn't seem to fit the standard values. Not only that, but when I did actually use that value in the formula I ended up with a cycle time of around 1989, and so rounding that up I would use the 2048 cycle value in bios. It didn't really help. I upped my mhz up a little more so that I can have a value closer to the 2048, and it still didn't help.

My refresh value is probably not 5.0uS like sandra tells me. It just doesn't seem correct (unless it really means 5.0ns; thus, meaning 7.8uS).

Also, how come there are TWO of the same cycle values in bios? There are TWO values of 2064, one towards the beginning and one towards the end. Is one for 256mb modules, and one for 512mb? I bet this question just ends up getting piled over, but more importantly El and I need to find out our REAL refresh values.

**EDIT**
P.S. I just bought 4 more TwinMOS SP at $198.XX shipped. After I get those I'll almost be at a total of 10gigs of TwinMOS UTT flowin through my house. ONLY FOUR MORE LEFT!!!

 

[HousERaT]

gotta love this stuff! tighter equals more stable. yeah I am going to give these sticks more juice 3.3 or 3.4 and see what that gets me. I have the weird timing/divider cuz I was having major cold boot issues from day one with three different bioses and I finally settled on that weirdness giving me the greatest shot at 2.8ghz dvd shrink stable/and gaming of course.

Basically the tref and other advice really helped me get super stable but I run memtest and see. also wouldn't mind running 1.5-2-2-5@240 but I think 3.4volts will be needed.

You should be able to hit 240 1.5-2-2-5 using 3.3v-3.4v. Some people believe running cas 1.5 gives no benefit but others believe that somehow it improves stability. I've not done any testing myself to see if there's a difference but I always run my memory with the tightest timings the system will allow. If you're still having problems with the cold boot issue you might want to take a look earlier in the thread about it. You might want to try the J17 mod if you have adjustable pots on your psu.

 

[g0dM@n]

...or you can just not power off your PSU. Use the front power switch!

Also, you can try RMAing your ram and getting a pair that won't have the issue.

I blame the RAM as well for this cold boot issue, as I had a TwinMOS pair where one dimm couldn't boot at 2.5v, and also an OCZ VX 4000 pair where one dimm couldn't boot at 2.5v.

The DFI board defaults to 2.5v BOOT when you kill the power to the board and then feed it power again. It bumps up to your bios vdimm reading after the system boots up.

Either replace the ram, or keep the power going to the board if you're not messing with any hardware.

I don't mind taking the chance myself, so when I have to move dimms around I still do it when the mobo is in standby mode.

 

[g0dM@n]

So I've been trying to figure out what other program other than Sisoft Sandra would tell me the refresh rate, so I FINALLY saw someone talk about "Everest" at DFI-Street.

Here it is... much better than that 5.0ns BS that Sandra gave me:

"Refresh Rate: Reduced (7.8us), Self-Refresh" (as it should have been in Sandra) Who knows what it says 5.0us in Sandra... if anything it should be 5.0ns.

 

[HousERaT]

gOdM@n you forgot to put up the Everest Benchmark screenies......

So are you now using the 7.8us and if so is it helping your overclock/stability with the ram?

 

[g0dM@n]

No, I'm not using the 7.8uS setting yet b/c I had to leave my house RIGHT AFTER I posted that.

My computer is still pretty stable at the 3120 setting, but my system did crash 2 nights ago. I gamed for about 3 hours (yes, I know... I actually had time that day), and my computer froze and repeated some sounds in the game, along with a scratchy, high-pitched squeaky sound. That is a problem I had way back on my NF7-S as well, and the problem went away when I removed my SB Audigy card. Well, I put in my SB Audigy card into this system (the DFI NF4 in my sig) less than two weeks ago. I'm assuming it's the hacked drivers that are causing the problem, but I have NO IDEA.

In any case, I will be testing out the 7.8uS setting when I have the chance... I did end up calculating what values I should be using in post #39.
The funny thing about that is that they all round up to the same tRef... I think 2048.

 

[felinusz]

Yes, for your specific refresh period, you will get "frequency brackets" where you're looking at the same tREF value. Scale your frequency up or down far enough, and your tREF will change.

About the two 2064 values, I'm not sure how they fit into the picture. Supposedly each refers to a different refresh interval, at a different frequency - of course, having two identical values that accomplish different things doesn't fit with the formula. That's part of why it's theoretical I suppose . I'd be calling it dodgy, if it didn't actually work for me . As it is, a lot of speculation is involved, nobody who has the definitve answer has come forward to share it as of yet.

Going through every single tREF value in the BIOS, while exhausting, really is worthwhile. If you find that the best value is the same one that the formula gives you, I think it adds some degree of credibility. I'm sold, but only because it worked with my sticks

 

[g0dM@n]

So I'm assuming that everyone would be using the 2048 or 2064 values pretty much. Aren't most of these 512 modules 7.8uS? I mean... what else COULD it be? I thought all 512 was 7.8; thus, giving most people the same tREF values b/c we're all around 245-260... if not, just a little higher or lower.

 

[zer0nix]

damn, all this trouble just to get 250fsb i'm beginning to rethink getting these (getting confused just reading this thread)...


minor topic steal what's a good, cheap source for jumpers?

 

[g0dM@n]

You're talking about the little tabs you use on jumpers? That's what you need?

 

[g0dM@n]

Btw, I memtested about a dozen loops on test #5 and a dozen loops on test #8 with these chips at 2-2-2-6, 200fsb 1:1 @ 1T, and default 2.6v... PASSED!

Right now I'm at my gf's, but I'm letting it do full memtest loops until I get home tomorrow night after the gym after work. $100 for 2-2-2 ram @ 2.6v... haha!

 

[HousERaT]

Btw, I memtested about a dozen loops on test #5 and a dozen loops on test #8 with these chips at 2-2-2-6, 200fsb 1:1 @ 1T, and default 2.6v... PASSED!

Right now I'm at my gf's, but I'm letting it do full memtest loops until I get home tomorrow night after the gym after work. $100 for 2-2-2 ram @ 2.6v... haha!

That is consistent with my findings. My sticks do run at 2.6v with 2-2-2-5 timings. Couldn't go any lower without a zillion errors.

 

[g0dM@n]

What do you mean lower? 1.5-2-2-5?

 

[felinusz]

No, he means a lower VDIMM voltage .

As an aside, I highly reccomend that you whip out the 5V jumper, and see how far your sticks will go with more voltage. Keep track of how they scale, stop when diminishing returns kick in if you are worried about damage. I would not be concerned with your sticks dying, so long as they are appropriately cooled. From what I've seen, these UTT BH memory ICS are a lot like BH-5 - they can take higher voltages, and often see gains from it.

 

[zer0nix]

You're talking about the little tabs you use on jumpers? That's what you need?

yep; haven't stepped foot at fry's since i posted here but sometimes their stuff is surprisingly overpriced...