Silver PI vs. Black PI gskill

[Soulkeeper]

what's the deal with the lower claimed voltage on the silver stuff ?
here's links:

Silver PI
Black PI

does it matter which One I get ?


thanks

 

[TC91]

Get the lower voltage 1100mhz stuff, it's very likely using better chips to achieve the higher clocks and at the same time at a very very low voltage for DDR2 1100mhz.

 

[BTRY B 529th FA BN]

Originally posted by: TC91
Get the lower voltage 1100mhz stuff, it's very likely using better chips to achieve the higher clocks and at the same time at a very very low voltage for DDR2 1100mhz.

+1

 

[Soulkeeper]

thanks, I just ordered it

hopefully 4 sticks will work without issue in my motherboard, not looking to overclock it
Running at the specified speeds would make me happy

 

[BTRY B 529th FA BN]

Put those in a UD3R or P board and that'll be a nice setup with that quad - see sig

 

[Soulkeeper]

I like my motherboard, this is my last bit of money (payed 30+ bucks more than I had originally planned, but I think I got the good stuff). so no recourse if it don't work. hopefully no DOA's

good to see it works in your gigabyte tho, that's a nice sign

 

[n7]

ProMOS versus PSC; you chose the better of the two. :thumbsup:

 

[Proximon]

My PI Blacks are great, but yeah I like to recommend that 1100 kit for folks going beyond 400 FSB with Intel.

 

[Soulkeeper]

I just got the memory and installed it about 30min ago
so far i'm extremely happy

running at 502 (1005) 5-5-5-15 1.8V 4x 2GB

compared to my 402 (804) 4-5-4-11 2.05v mushkin 2x 1GB

18% more mem bandwidth
13% lower voltage
4x the capacity

Also I was never able to run other mem dividers with my mushkin (and it wouldn't overclock very well), it only did 1:1 stabily even when underclocked, this new memory accepts them fine

I havn't tryed anything higher yet (leaving cpu/fsb at the same speeds as before), just stress testing.

Pic 1
Pic 2
Pic 3
Pic 4

 

[BTRY B 529th FA BN]

jealous! i love that chair!

 

[Soulkeeper]

haha

 

[BTRY B 529th FA BN]

time to change your sig

 

[Soulkeeper]

There we go

took the memory out of this system and plopped it in my c2d webserver, which had 512MB of pc2-4200

so both my system have quadruppled the memory now

one of the best upgrades i've ever made

Got me a little box of memories here .....

 

[n7]

Bah, what's with the underclocking.

Raise the ratio & see how far OVER 1100 it can do...also, i'd be really curious to see how far it can go CAS 4.

 

[Soulkeeper]

I'll do that tomarrow
gonna run it over night and make sure it's fully stable first

I'll let you know what it hits, maybe post some benchmarks too
the next ratio up puts it at exactely 1200 btw

 

[lopri]

Those are some real nice sticks. Had I not already have million DDR2 sticks I'd have bought them in a heartbeat. Enjoy!

 

[Soulkeeper]

Ok I just tryed:

800 @ 4-4-4-12 1.9v -stable
1000 @ 4-4-4-12 2.0v -no post memory error beep
1200 @ 5-5-5-15 1.95v -partial post then lockup
1200 @ 5-5-5-15 2.0v -partial post then lockup
1200 @ 6-6-6-18 2.0v -partial post then lockup

1005 @ 5-5-5-15 1.8V -stable thru the night, but system hardlocked at about 11am
now trying this setting at 1.92v and a slight boost to chipset voltages for insurance.

looks like there is no real overclocking headroom for me

note: this is a workstation/server so i'm likely to not try any other messing around
my main priority is getting this thing 100% stable, and i'm a wuss with the voltages anymore (not like I used to be)

regardless i'm happy with what I got

 

[Soulkeeper]

allocating 1200MB a peice (2.4GB total) two zero-initialized memory blocks
Time: 3840.805033MB/s
change all values of allocated memory 1 element at a time
Time: 2507.197747MB/s
copy array a to array b
Time: 2975.253330MB/s
copy array a to array b in incrementing steps of 32MB
Time: 2988.598547MB/s
copy array a to array b in incrementing steps of 128MB
Time: 2988.580593MB/s
copy array a to array b 1 element at a time (dumb test)
Time: 1947.587183MB/s
free arrays
Time: 0.094334
Done.




heres some results from a program I wrote earlier today
note those are combinations of write/write read/write so doubling those numbers is essentially my mem bandwidth

 

[daw123]

Originally posted by: Soulkeeper
There we go

took the memory out of this system and plopped it in my c2d webserver, which had 512MB of pc2-4200

so both my system have quadruppled the memory now

one of the best upgrades i've ever made

Got me a little box of memories here .....

Geez, are you collecting processors?

 

[Blain]

I can touch 8 CPU's on the desk in front of me from chair.
There are also 6 HDs within arms reach.

AT members tend to gather quite a collection of parts over time.

 

[n7]

Originally posted by: Soulkeeper
Ok I just tryed:

800 @ 4-4-4-12 1.9v -stable
1000 @ 4-4-4-12 2.0v -no post memory error beep
1200 @ 5-5-5-15 1.95v -partial post then lockup
1200 @ 5-5-5-15 2.0v -partial post then lockup
1200 @ 6-6-6-18 2.0v -partial post then lockup

1005 @ 5-5-5-15 1.8V -stable thru the night, but system hardlocked at about 11am
now trying this setting at 1.92v and a slight boost to chipset voltages for insurance.

looks like there is no real overclocking headroom for me

note: this is a workstation/server so i'm likely to not try any other messing around
my main priority is getting this thing 100% stable, and i'm a wuss with the voltages anymore (not like I used to be)

regardless i'm happy with what I got

I'd suggest working with one ratio when OCing till FSB gets too high; then switch to the next ratio.

Otherwise, you're really just switching ratios, which isn't going to show much, as in your case.

I generally lower the multiplier to below stock to make sure the CPU is removed from the equation, & then bump up FSB & loosen timings along the way.

So as an example:
If 7x333 5:6 DDR2-800 4-4-4 is stable, next try 7x340 DDR2-816 4-4-4, & keep bumping till it doesn't POST, or till Memtest86+ test 5 fails.
That specific test is really good for quick testing.
For serious testing, run the full Memtest86+ pass, or better, get into Windows & run HCI Memtest for a while.
Memtest86+ isn't very good compared to HCI Memtest for checking overclocked stability, especially with larger amounts of RAM...

When 4-4-4 isn't stable anymore, switch to 5-4-4 & see how far, then 5-5-4, then 5-5-5, 6-5-5, etc...

Just how i do it.

I haven't done really thorough testing of my RAM for a while, but for 2x2 (4x2 is lower) i can tell you it does roughly DDR2-825 4-4-3, DDR2-900 5-4-3, & DDR2-1100 5-5-5.
Haven't really tested only 2x2 for CAS 6, & it's impossible fighting with all 8 GB at high speeds, so i don't generally bother.

 

[Soulkeeper]

yeah i've done all that fun stuff before, Just seems like I don't have the passion for it anymore
wanted to keep my fsb constant basically (didn't bother moving it at all here)

getting an extra 100MHz out of my memory won't make any difference to me, specially if I have to downclock my cpu any

Now if I got payed to review things then i'd have lots of passion