ATT: Old skool folks - BH-5 related question

[[XC] Oj101]

Hey guys, I know this isn't really the correct section but I think because my question is about extreme RAM this is more appropriate than General Memory. When you were running BH-5, what was the best Everest latency you managed? I THINK I've just seen my first picture of DDR2/3 where the latency is better than the best I ever managed. Could you please share some screenshots?

[BenchZowner]

Wish I could remember :D
Depends on the platform as well, AMD ? Intel ?
All I remember is Corsair & GeIL lovin' at 4V DDR-500+ 2-2-2-5 :D

[[XC] Oj101]

DDR500? That's a touch low, you mean closer to DDR600 :D I forgot to mention I'm only interested in the AMD results.

[kiwi]

Don't have screens but 250-260 2225 with overclocked AMD cpu on air will get you around 35ns in everest. However, that was an old everest version, they corrected latency measuring algorithm in later versions and latency is being reported slower.


Yea, 250-260 was piece of cake. 280+ is more interesting :D

[[XC] Oj101]

35ns is waaaaaay to slow, I was in the high 20s and I know I wasn't the fastest.

[BenchZowner]

Those latencies were incorrect though :D
If I recall correctly I've seen 18ns or so back then.

[[XC] Oj101]

I'm not talking about "bugged" runs due to an ancient version of Everest, I did mine in the last year of so.

[[XC] Oj101]

All I can find of my old results is this:
http://i140.photobucket.com/albums/r35/Oj10101/2000200.jpg

[K404]

4V DDR-500+ 2-2-2-5 :D

Sometimes less is more y'know.... :p: 3.2-3.35v was enough for 260-265 on my Redlines. Memories!

3D03 WR was 16K with a 6800GT.... just before SLI came out

[BenchZowner]

I said DDR-500+.

My personal best was 280ish MHz ( 5MHz up/down ) with the Corsair XMS 3200C2something BH-5 at 3.8-3.9Vish with an Opteron 146 on a DFI LP nF4 of course.
Good memories :)

[K404]

Ah.... 500+++ or 550+ :p: :D

[DuraN]

I think my Kingston HyperX3500 BH-5 ran 250+ at 2-2-2-5 but back then people weren't using Everest :D

Good times indeed, I still have memories of my NF7-S 2.0...

[boblemagnifique]

Old Screen (March 2006) BH5 + Caskade 2 stages :D

http://membres.multimania.fr/boblemagnifique/Overclocking/Bh5/kit%20dual/lat318.jpg

[[XC] Oj101]

F me that's fast :(

Thanks for the latency! DDR3 is officially faster AT LAST! :D

[ImportantAwareness89]

Oh the good old days. My OCZ VX used to do like 265 2-2-2-5 or something. I don't know if they fried but I ran them around 250 2-2-2-5 (rated speeds) consistently and I got a lot of OS issues. I replaced my HD once and it kept happening. Every 1-2 months my OS would corrupt up. I finally swapped out RAM even though I memtested perfectly fine for 24 hrs+ and I even Primed 36 hrs stable. Switched to some other non BH-5 insane voltage stuff (TCCDs was it?) and I was fine from then on.

[M.Beier]

I think I had 28.7ns, but cant find sh1t....
Was ~283mhz 2-2-2-5, 2x512MB blocks

These are the ones;
http://img115.imageshack.us/img115/4489/img2006.jpg

EDIT;
Funny thing; I remember it was 3.93v, because DFI nF4 Ultra-D didnt like 4v and there was the +0,03v option on the board, heh...

[[XC] Oj101]

28.7ns - I'm feeling it ;)

[EnJoY]

This is an old screenshot from when I sold my last set of BH5. Mushkin LVL2 Black PC3500 2x512MB.

http://www.techreaction.net/wp-content/uploads/2010/05/10x270_bh5sale.gif

Mmmmmmmm :slobber:

[[XC] Oj101]

Considering everything else, that's a nice read preamble but I'm sure you could've done Trtw 1? Your picture is missing the one crucial aspect: Everest latency benches :D

[M3kk]

Oh, nice old times :).
I loved the DFI board.

Sorry, no Everest..

http://mekk.homeip.net/imagesharing/uploads/bd977c79ed.jpg

http://mekk.homeip.net/imagesharing/uploads/30da17b156.jpg

http://mekk.homeip.net/imagesharing/uploads/fc0a9d59c7.jpg

http://forum.crazypc.ro/attachment.php?attachmentid=11072&d=1144234795

http://mekk.homeip.net/imagesharing/uploads/24d810f186.gif

http://mekk.homeip.net/imagesharing/uploads/edd9eb18df.gif

[EnJoY]

Considering everything else, that's a nice read preamble but I'm sure you could've done Trtw 1? Your picture is missing the one crucial aspect: Everest latency benches :D

I don't really remember, but I'm fairly sure TRTW couldn't be left at 1 after 255-260MHz or so. It needed to be upped a notch. Again, I don't really remember. :)

[WeStSiDePLaYa]

BH-5 were some fun days. Mushkin BH-5 originals were killing it.

[ramfan]

this is all i can find
ut nF3 250GB with my custom vdimm power supply
corsair pc3500 plat

http://i256.photobucket.com/albums/hh199/redhat_ownage/270.jpg
http://i256.photobucket.com/albums/hh199/redhat_ownage/CSC_2292.jpg

[WeStSiDePLaYa]

this is all i can find
ut nF3 250GB with my custom vdimm power supply
corsair pc3500 plat



Damn, that made me forget that i had modded my nf3 250gb the same way.

Still have that board, some bh-5 and an old sempron. might have to get it out and have some fun.

[[XC] Oj101]

I can't make out what you've done, but 5V rail -> pot -> DIMM slot?

[ramfan]

I can't make out what you've done, but 5V rail -> pot -> DIMM slot?

5VSB rail -> 2x 2200uF caps -> 30V 10 amp mosfet -> dimm
its controlled by a modded oskar wu beta bios

but i killed the chipset trying for the HWbot skt 754/939 htt contest

i still have that device i made if anyone wants it just ask
or if anyone has any nf3 250gb chipsets laying around i can resurrect my dfi

[[XC] Oj101]

Do you have a link to the BIOS and schematics? That would be greatly appreciated! :)

OT: Would my idea work?

[ramfan]

Do you have a link to the BIOS and schematics? That would be greatly appreciated! :)

OT: Would my idea work?

bios
http://www.overclockers.com/forums/showthread.php?t=328318

DFI beta 10/15 - 4.0 vdimm - by Jess1313 & Samurai Jack(aka JSJ) (http://webpages.charter.net/jess1313/DFI%20754%20Beta%20Bios/Beta%20A15/JSJ-A15.ZIP)

http://i256.photobucket.com/albums/hh199/redhat_ownage/CSC_2288-2.jpg

the tricky part is desoldering the mosfet leg and lifting it so you can solder the wire from the external mosfet to the board
the 5VSB must be used or the board will not turn on


there is an ez way to do a 4v mod on the nf3 but you need heatsinks on the vdimm mosfets
just connect where the red line is
http://i256.photobucket.com/albums/hh199/redhat_ownage/CSC_2319.jpg

but you will need the heatsinks on mosfets Q38 and Q50
dont worry about VTT it scales with the vdimm

your idea would not work the pot would just burn up

[[XC] Oj101]

Thanks mate :D

[ramfan]

no problem
protip: if you know anyone that can mod a bios, it has options over 4V hidden inside

[[XC] Oj101]

Tictac/Bigtoe? :D

[George_o/c]

protip: if you know anyone that can mod a bios, it has options over 4V hidden inside

Didn't know that, after all these years! Now that's interesting :yepp:

[[XC] Oj101]

I'm with you on that one, it's been over four years since anyone other than a tiny handful have cared about the boards and it only comes out now :(

[WeStSiDePLaYa]

bios
http://www.overclockers.com/forums/showthread.php?t=328318

DFI beta 10/15 - 4.0 vdimm - by Jess1313 & Samurai Jack(aka JSJ) (http://webpages.charter.net/jess1313/DFI%20754%20Beta%20Bios/Beta%20A15/JSJ-A15.ZIP)



the tricky part is desoldering the mosfet leg and lifting it so you can solder the wire from the external mosfet to the board
the 5VSB must be used or the board will not turn on


there is an ez way to do a 4v mod on the nf3 but you need heatsinks on the vdimm mosfets
just connect where the red line is
http://i256.photobucket.com/albums/hh199/redhat_ownage/CSC_2319.jpg

but you will need the heatsinks on mosfets Q38 and Q50
dont worry about VTT it scales with the vdimm

your idea would not work the pot would just burn up

I never used the 5vsb for my mod. Just regular 5V line from the atx. Maybe yours didnt boot because of a psu issue?

I used 5V, probably would have cleaner power and has more juice than 5vsb.

You probably couldnt boot because you have too much capacitance. 4000uf is a lot.

If I remember right i went with two 470 ultra low esr caps.

Quality capacitance > bulk capacitance

[Lastviking]

Still got all my good bh5 sticks and my older brother is using my Dfi Nf4 Venus + 4x512mb bh5 24/7

I think i got..

6x256mb bh5
4x512mb bh5

To bad i have forgot most of the settings...

[jmax_oc]

Hehe, very nice to see such topic. I did 29,7ns with 2x512Mo at 289MHz 2-2-2-5. I will try to get these old screens.

[knopflerbruce]

This 4v mod sounds interesting... does it give instant 4v, or does it just enable youto select higher vdimm (if you use a BIOS which lists 3.3v++)

[[XC] Oj101]

You typed that message in French and used an online trnslator :p:

It's very nice to see you talking about DDR400 again! Well I guess that pretty much settles it, I recently saw DDR3 doing 28.5ns on 24/7 clocks which is the first time I've seen sub 30ns since BH-5. I didn't think we'd see it again :)

[[XC] Oj101]

The latter if I'm understanding it correctly. Getting a solid 4v is easy enough without any BIOS modifications.

[jmax_oc]

Yes 4V is ok without BIOS mod. BIOS mod give you +0.04V option.
I don't save the latency score but it was 29.9ns at 290MHz 2-2-2-5 with 2x512 sticks.

Here are some screens for BH5 overclocking :
http://img440.imageshack.us/img440/3499/maxspi32mwinnie3hp.jpg

http://img176.imageshack.us/img176/7637/dcp11356ph.jpg

http://img199.imageshack.us/img199/838/dcp11264gh.jpg

http://img345.imageshack.us/img345/6465/dcp11212ga.jpg

http://collabo.free.fr/overclock/jmax/memory-2700.JPG

[ramfan]

I never used the 5vsb for my mod. Just regular 5V line from the atx. Maybe yours didnt boot because of a psu issue?

I used 5V, probably would have cleaner power and has more juice than 5vsb.

You probably couldnt boot because you have too much capacitance. 4000uf is a lot.

If I remember right i went with two 470 ultra low esr caps.

Quality capacitance > bulk capacitance

if you turn off the psu and clear the cmos the board will not turn back on if you use the 5V rail i had this issue come up a few times.
also you probably did not do the mod the way i did, bypassing the boards vdimm power supply


This 4v mod sounds interesting... does it give instant 4v, or does it just enable youto select higher vdimm (if you use a BIOS which lists 3.3v++)

it works just like the 4V jumper on the dfi nf4 boards, only there is no jumper and you have to flash the bios.

[M.Beier]

Hehe, very nice to see such topic. I did 29,7ns with 2x512Mo at 289MHz 2-2-2-5. I will try to get these old screens.

Are you sure buddy? Should be lower, mine was just made with a 3700+ SD.... Might have been the subbies that were a tad higher then, I used 01SE-optimized subtimings for low latency.... (As most remember; Not same as highest bandwidth settings)

[WeStSiDePLaYa]

if you turn off the psu and clear the cmos the board will not turn back on if you use the 5V rail i had this issue come up a few times.
also you probably did not do the mod the way i did, bypassing the boards vdimm power supply



it works just like the 4V jumper on the dfi nf4 boards, only there is no jumper and you have to flash the bios.

Nope, same mod. I remember hearing when it first came out that too much bulk capacitance would cause cold boot issues. because bulk capacitance causes a delay because the caps need to charge up.

Which would be the problem you encountered using 5v, which isnt always active unlike 5vsb.

Also, your board could have died because using the the 5vsb would be feeding the vdimm at ALL times unless the powersupply was turned off.

Which would also make it dangerous for the ram.

[jmax_oc]

Are you sure buddy? Should be lower, mine was just made with a 3700+ SD.... Might have been the subbies that were a tad higher then, I used 01SE-optimized subtimings for low latency.... (As most remember; Not same as highest bandwidth settings)
Yep sure, unfortunately :shrug:

[ramfan]

Nope, same mod. I remember hearing when it first came out that too much bulk capacitance would cause cold boot issues. because bulk capacitance causes a delay because the caps need to charge up.

Which would be the problem you encountered using 5v, which isnt always active unlike 5vsb.

Also, your board could have died because using the the 5vsb would be feeding the vdimm at ALL times unless the powersupply was turned off.

Which would also make it dangerous for the ram.

the nf3 originally gets its vdimm power from the 5vsb rail trust me ive checked the traces all i did was bypass the crap power supply and built my own.
if you plug in your nf3 you will see the vdimm light come on before you turn on the board.

[[XC] Oj101]

330MHz... IN MEMTEST??? What cooling were you using there??

[George_o/c]

Holy :banana::banana::banana::banana: dude! And I thought 280MHz 2-2-2-5 was good :o Look at that 320MHz run :eek2:

[M.Beier]

Holy :banana::banana::banana::banana: dude! And I thought 280MHz 2-2-2-5 was good :o Look at that 320MHz run :eek2:

for 2x512MB its sick, for 2x256MB, its ok, but not that cool :-P
I think jmax or lastviking, cant remember who, is only to break 300mhz on 2x512MB

[Barr3l Rid3r]

so much fun w/ them:

http://i42.tinypic.com/1h76vo.png


http://i42.tinypic.com/1118rqh.jpg


http://i43.tinypic.com/2yk1jly.jpg

[eBoy0]

Anyone remember
http://www.rostec.co.il/tep_catalog/images/booster.jpg
That + Redline = Pwnage :)
Or some dead ram/dimm slots

[jmax_oc]

for 2x512MB its sick, for 2x256MB, its ok, but not that cool :-P
I think jmax or lastviking, cant remember who, is only to break 300mhz on 2x512MB
LV did it :)


330MHz... IN MEMTEST??? What cooling were you using there??
LN2 (-70°C / -110°C). If you want to see more details / pics : http://www.jmax-hardware.com/forum/index.php/topic,78.0.html

[[XC] Oj101]

Anyone remember
http://www.rostec.co.il/tep_catalog/images/booster.jpg
That + Redline = Pwnage :)
Or some dead ram/dimm slots

I bought one last year :D

[knopflerbruce]

LV did it :)


LN2 (-70°C / -110°C). If you want to see more details / pics : http://www.jmax-hardware.com/forum/index.php/topic,78.0.html

How much do you earn on LN2-ing BH5? 30MHz @ 2-2-2-5?

[WeStSiDePLaYa]

the nf3 originally gets its vdimm power from the 5vsb rail trust me ive checked the traces all i did was bypass the crap power supply and built my own.
if you plug in your nf3 you will see the vdimm light come on before you turn on the board.

No. you are wrong. 3.3v rail was used. This is common knowledge.

To further illustrate that, if 5vsb was originally used, then this mod would have not been necessary. as you are replacing the original 3.3v supplied circuit, with a 5vsb circuit. the layout of the circuit is essentially the same otherwise.

Also, read this thread. It uses 3.3v.

http://forums.techpowerup.com/showthread.php?t=4751

Also, here is a thread where the creator of the mod talks about the original vdimm source being 3.3v.

http://www.xtremesystems.org/forums/showthread.php?t=70566&highlight=nf3+250gb+vdimm+mod

[George_o/c]

for 2x512MB its sick, for 2x256MB, its ok, but not that cool :-P
I think jmax or lastviking, cant remember who, is only to break 300mhz on 2x512MB

Marc, 'still I think it's impressive! :up:
Got me an Abit IC7-G, will have it modded asap. Then I just need to find again a good BH5 kit, and I'll continue benching S478 CPUs :D

BTW, if anyone has a good kit, PM me :cool:

[ex2cib]

I almost forgot about everest. Never had bh5 myself, but the best latency i ever had was 32.4ns with a 1gb set of the "bad" tcc5 (2.5-3-3-7 @ 322MHz). guess that wasn't too bad for a 3000+ @ 2900.

thinking back, i just wish i wouldve learned how to clock them before my cpu started to die:shrug:

[ramfan]

No. you are wrong. 3.3v rail was used. This is common knowledge.

To further illustrate that, if 5vsb was originally used, then this mod would have not been necessary. as you are replacing the original 3.3v supplied circuit, with a 5vsb circuit. the layout of the circuit is essentially the same otherwise.

Also, read this thread. It uses 3.3v.

http://forums.techpowerup.com/showthread.php?t=4751

Also, here is a thread where the creator of the mod talks about the original vdimm source being 3.3v.

http://www.xtremesystems.org/forums/showthread.php?t=70566&highlight=nf3+250gb+vdimm+mod

im sorry but these sources have no information of the actual workings of the board, check for yourself if you want

http://i256.photobucket.com/albums/hh199/redhat_ownage/overview.jpg

[Viss]

Loved those mems :D

http://img.photobucket.com/albums/v292/Viss1/Mushkin2x512.jpg

http://img.photobucket.com/albums/v292/Viss1/sandra280.jpg

http://img.photobucket.com/albums/v292/Viss1/1M3.jpg

http://img.photobucket.com/albums/v292/Viss1/superpi3250.jpg

And OCZ VX (CH5):
http://img.photobucket.com/albums/v292/Viss1/VX275bandwidth1024.jpg

[felix_w]

I bought one last year :D

I got it as gift from a retailer in Greece..now works on a P4C800-E Dlx...:up:
Torturing a set of Kingston 512MB sticks with BH5 that i found ...

[[XC] Oj101]

Motherboard Monitor? That brings back fond memories :)

[Viss]

Stil have 2x256 BH5 and the OCZ VX, PM me if u can use them Oj101

[jmax_oc]

How much do you earn on LN2-ing BH5? 30MHz @ 2-2-2-5?
Yes 30MHz. But I make only one LN2 run and it was my first one without any help. I think that 340MHz 1M 350MHz screen is possible with good kit, supply mod (low ESR cap) and modern skill :)

[Cecil]

Spent the last few days trying to find some 2x256 BH-5s. Think these are them?
http://www.zipzoomfly.com/jsp/ProductDetail.jsp?ProductCode=85005-1
Not sure Id want to pay $60 for them though. I have a set of OCZ gold 512s, but want the 256s.

[proxis]

doubt youll find any brand new winbond these days, those may as well be infineon

[Cecil]

http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&item=150440226541&ssPageName=STRK:MEWAX:IT

How about those? I put in an offer. They said they are exactly as pictured, cause I asked about the stickers having so little info on them.

[Lastviking]

http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&item=150440226541&ssPageName=STRK:MEWAX:IT

How about those? I put in an offer. They said they are exactly as pictured, cause I asked about the stickers having so little info on them.

If i dont remember wrong EL was not any great bh5 sticks.

[Cecil]

OCZ EL DDR PC3200 Gold = Winbond UTT BH5
OCZ EL DDR PC3200 Gold C2 = Winbond UTT BH5
OCZ EL DDR PC3200 Gold C2 Rev2 = Winbond UTT BH5
http://whirlpool.net.au/wiki/?tag=rmp_rsc

I figure they have a good shot at being BH-5. Im thinking they are probably early samples, or possibly review samples that didnt have the part number and timings on them. For $20, its worth trying.

[knopflerbruce]

http://whirlpool.net.au/wiki/?tag=rmp_rsc

I figure they have a good shot at being BH-5. Im thinking they are probably early samples, or possibly review samples that didnt have the part number and timings on them. For $20, its worth trying.

http://cgi.ebay.com/Mushkin-DDR-PC3200-Dual-Channel-Kit-512MB-BH-5-Chipset-/120568996282?cmd=ViewItem&pt=LH_DefaultDomain_0&hash=item1c1278e1ba

[Cecil]

http://cgi.ebay.com/Mushkin-DDR-PC3200-Dual-Channel-Kit-512MB-BH-5-Chipset-/120568996282?cmd=ViewItem&pt=LH_DefaultDomain_0&hash=item1c1278e1ba

I messaged him about them, but havent got a responce. Those came in BH-5 and BH-6.

I got an offer for a set of CMX256A-3200LLPT on another forum for $40, so Im grabbing them too just incase the OCZ golds arent.

[WeStSiDePLaYa]

im sorry but these sources have no information of the actual workings of the board, check for yourself if you want

http://i256.photobucket.com/albums/hh199/redhat_ownage/overview.jpg


you can ignore reason all you waast. but it doesnt chsnge the fact that if 5vsb was used as the vdimmm source, the mod to change vdimm source from 3.3v to 5v wouldnt exist.


you have to realize youre wrong, and it probably the reason your board died.

[jmax_oc]

Hmmm, in 95% of mainboards, RAM power come from +3.3V but Ramfan is right.

In DFI BIOS and in '3.2V mode' you get power from 3.3V. When you switch to 4V mode, the power come from +5V (power can't come from 3.3V)

[M.Beier]

Hmmm, in 95% of mainboards, RAM power come from +3.3V but Ramfan is right.

In DFI BIOS and in '3.2V mode' you get power from 3.3V. When you switch to 4V mode, the power come from +5V (power can't come from 3.3V)

But we all used DFI nF4 for ram clocking back then :D

[jmax_oc]

But we all used DFI nF4 for ram clocking back then :D

Of course :D

[ramfan]

you can ignore reason all you waast. but it doesnt chsnge the fact that if 5vsb was used as the vdimmm source, the mod to change vdimm source from 3.3v to 5v wouldnt exist.


you have to realize youre wrong, and it probably the reason your board died.

lol my board died because i put 1.9V into the NB trying to break 543mhz HTT.
you can argue all you want but im right :rofl:

Edit: someone send me a nf3 250gb northbridge chip so i can fix my board and school this kid

[ramfan]

pics or it didnt happen right?
conditions: plugged the 24pin in only and turned on the psu board is off nuff said

http://i256.photobucket.com/albums/hh199/redhat_ownage/SSC_2727.jpg
http://gamerofsorts.com/wp-content/uploads/2010/02/stfu_noob.jpg
http://www.youtube.com/watch?v=Qg7jA-H-jMo

[[XC] Oj101]

but it doesnt chsnge the fact that if 5vsb was used as the vdimmm source, the mod to change vdimm source from 3.3v to 5v wouldnt exist.

Why exactly not?:shrug:

[WeStSiDePLaYa]

Why exactly not?:shrug:

Because then the options upto 4v in bios would have already been available.

and DID YOU READ this thread or not?

http://forums.techpowerup.com/showthread.php?t=4751

Also, To FURTHER support my claim, many people who DIDNT hardmod their boards were able to get higher vdimm by raising the 3.3v rail. and the max vdimm would always be about .2v lower than 3.3v

I PERSONALLY tested this on a RUNNING board with an antec true control psu.

Lowering the 3.3.v rail to 3.1v would provide vdimm at 2.9v raising the 3.3v rail to 3.5v would raise vdimm to 3.3v WHILE the board was running, no changes in bios.

Also, have you thought about how many psu's have a only 2 or 3 amps on the 5vsb rail? which gives you 20 to 30 watts. and each stick of ddr consumes about 10 watts? and thats not overclocked or overvolted.


Also, and one last thing, show me where it says that is for vdimm source voltage.


Oh, and just for you to have. Here is information from Oskar_Wu himself, the designer of the board, Stating that with the 4.0v bios, the max vdimm is .1-.2v lower than the 3.3v.

WHY? because vdimm is derived from the 3.3v rail.

Are you going to argue with the man who designed the entire board? or are you going to realize maybe where you are measuring doesnt feed the vdimm?


6. Regarding beta bios with 3.3 and 4.0v vdimm options at http://www.ocforums.com/showthread.php?t=328318. From Oskar_Wu designer of this DFI board regarding modded bios he released for >3.1v vdimm

Criteria for usage for running higher VDIMM > 3.1V

1 . CPU VID control must set to voltage other than Auto
2 . Your ATX 3.3V must be higher than your target highest VDIMM for about 0.1~0.2V
Ex : 3.3V ~ 3.4V VDIMM with ATX 3.3V@3.5V~3.6V is possible


If DFI Street hadnt shut down, I could post the exact quotes I read from Oskar_Wu on this very issue.

[Cecil]

Ive been playing with my 2x512 OCZ golds a little, and was able to get 515Mhz 2-2-2-5-7 with 3.6vdimm on a SLI-DR.

A few posts back I posted I got the 2x256 golds that had a wierd sticker on them. Well, turns out they are fake ram. The PCB slides right out of the "heatspreader" and there are no chips on the PCB at all :rofl:
And that was from SVC :down:

I did get a set of 2x256 corsairs and they for sure are BH-5, checked the ICs myself. What sucks now, is my SLI-DR seems to have a bad dimm slot. Two sticks of them, or my TCCD wont post. But with both of my 2x512 golds, it does :confused:

[Nanometer]

OCZ 4000VX 260MHz with 2-2-2-6 on 3.4 volts... ahh the good days

[[XC] Oj101]

Because then the options upto 4v in bios would have already been available.

and DID YOU READ this thread or not?

http://forums.techpowerup.com/showthread.php?t=4751

Also, To FURTHER support my claim, many people who DIDNT hardmod their boards were able to get higher vdimm by raising the 3.3v rail. and the max vdimm would always be about .2v lower than 3.3v

I PERSONALLY tested this on a RUNNING board with an antec true control psu.

Lowering the 3.3.v rail to 3.1v would provide vdimm at 2.9v raising the 3.3v rail to 3.5v would raise vdimm to 3.3v WHILE the board was running, no changes in bios.

Also, have you thought about how many psu's have a only 2 or 3 amps on the 5vsb rail? which gives you 20 to 30 watts. and each stick of ddr consumes about 10 watts? and thats not overclocked or overvolted.


Also, and one last thing, show me where it says that is for vdimm source voltage.


Oh, and just for you to have. Here is information from Oskar_Wu himself, the designer of the board, Stating that with the 4.0v bios, the max vdimm is .1-.2v lower than the 3.3v.

WHY? because vdimm is derived from the 3.3v rail.

Are you going to argue with the man who designed the entire board? or are you going to realize maybe where you are measuring doesnt feed the vdimm?



If DFI Street hadnt shut down, I could post the exact quotes I read from Oskar_Wu on this very issue.

Thank you for getting aggressive in answering my question :rolleyes: But it doesn't tell me why if the 5vsb voltage were used it would not allow a 3.3v to 5v change? One the DFI board (as far as I can recall) the jumpers changed the rail in use (hence the vdimm regulators getting hotter when using the 5v source with vdimm below what the 3.3v rail can supply compared to using the 3.3v rail for the same voltage). If the rail switch is not only possible but actually used in practice, why would a mod to change from one rail to another on board pre NF4 be impossible if you were to mod it to use the 5vsb rail? And regarding the power consumption being greater than the rail can provide, how did ramfan get his to work?

Try not to flame when you answer this :shakes:

EDIT: After a bit of Googling I'm seeing that people COULDN'T get higher voltages by adjusting their rails. I'm not arguing, when I never tried it myself. I'm just looking for answers :)

[[XC] Oj101]

My DFI NF4 is also sad :(


DFI NForce 4 Ultra-D - this board still works after being baked several times, one working DIMM slot, one working PCIe 16, other works at x1, no PCI/LAN/sound/SATA/IDE. It has spent in the region of 7000 hours subzero and at least ten hours sub zero without any insulation so it lasted me well

[Lastviking]

Thank you for getting aggressive in answering my question :rolleyes: But it doesn't tell me why if the 5vsb voltage were used it would not allow a 3.3v to 5v change? One the DFI board (as far as I can recall) the jumpers changed the rail in use (hence the vdimm regulators getting hotter when using the 5v source with vdimm below what the 3.3v rail can supply compared to using the 3.3v rail for the same voltage). If the rail switch is not only possible but actually used in practice, why would a mod to change from one rail to another on board pre NF4 be impossible if you were to mod it to use the 5vsb rail? And regarding the power consumption being greater than the rail can provide, how did ramfan get his to work?

Try not to flame when you answer this :shakes:

EDIT: After a bit of Googling I'm seeing that people COULDN'T get higher voltages by adjusting their rails. I'm not arguing, when I never tried it myself. I'm just looking for answers :)

I dont have used any Dfi Nf3...but i have used Dfi Nf2 and Nf4, with the Nf2 board i tweaked my +3.3v line to about 3.7-3.8v to feed the dimms with more than +3.3v(vdimm modd also to tweak the vdimm higher).. But vdimms was feeded by +3.3v on nf2

I think the Nf4 did have +3.3v feed and +5v feed(change with jumper)

[karbonkid]

BH5?? pffffttt. TCCD ftw

you guys are making me want to run back home and fire up my nf3-250gb, mobile athlon64 3700 (newark) and gskill TCCDs =D

ahh, memories, good memories...

[[XC] Oj101]

Let me just clarify that: EDIT: After a bit of Googling I'm seeing that people COULDN'T get higher voltages (on their NF4) by adjusting their rails.

My G.Skill TCCDs are in my sig, and again here :D 2x256MB G.Skill TCCD @ 350MHz 3-4-4-8 3.1v (http://valid.x86-secret.com/show_oc.php?id=288119). BH-5 FTW though, show me TCCDs running 1.5-0-0-0 :rofl:

[karbonkid]

My G.Skill TCCDs are in my sig, and again here :D 2x256MB G.Skill TCCD @ 350MHz 3-4-4-8 3.1v (http://valid.x86-secret.com/show_oc.php?id=288119)

Is that a challenge? :cool: I seem to remember (my memory is fuzzy) my 2x256 did 348 3-4-4-8 but I reckon they'll go further. Ans 3.1v is really, really high for tccd! Mine didn't like anything over 2.9. What drive strength you running at?


BH-5 FTW though, show me TCCDs running 1.5-0-0-0 :rofl:

True, true. Remember GEiL One? They had both TCCD and BH-5 versions, binned for CAS 1...

The only thing I have against BH5 is the high VDIMM eventually killing the CPU's IMC... Lost 2 semprons that way...

[MrIcee]

Hehehe:)

Oj101, I don't know about latencies with the BH-5 I used to run as we didn't run Everest like we do today, we were using Sandra exclusively. But here are a couple links of what some OCZ PC3500 stuff did for me in the single channel A64 days on the Shuttle board. PS...don't ask about volts:D

http://www.ocztechnology.com/aboutocz/press/2003/78

Sandra results: http://www.ocztechnology.com/products/hall_of_fame/page_13.html

http://img198.imageshack.us/img198/1239/mricee310b.jpg

310Mhz 1-1 at that time I believe was a first on the AMD platform. THOSE were certainly the days:)

Best regards,

Randi:D

[[XC] Oj101]

Yip, I remember them. I also remember anything below CAS2 being purely academic :p: 348MHz, damn that's close and I can't even play along as they later died :( Either 3.7v ( :rolleyes: ) or drive strengths killed them, I honestly think it was the latter. As for the IMC story, well when I was still on air cooling I ran 3.7vdimm with 1.55vcore for WELL over a year.

[[XC] Oj101]

MrIcee, with such a high HT speed you must've been on 754/939 (A64). On A64 the memory speed is derived from the CPU speed, not the HT speed - eg. CPU/11. There are no non-integer dividers, they're all positive integer numbers so you would've been rounded to CPU/6 or 284.63MHz :( Correct me if I'm wrong.

[WeStSiDePLaYa]

Thank you for getting aggressive in answering my question :rolleyes: But it doesn't tell me why if the 5vsb voltage were used it would not allow a 3.3v to 5v change? One the DFI board (as far as I can recall) the jumpers changed the rail in use (hence the vdimm regulators getting hotter when using the 5v source with vdimm below what the 3.3v rail can supply compared to using the 3.3v rail for the same voltage). If the rail switch is not only possible but actually used in practice, why would a mod to change from one rail to another on board pre NF4 be impossible if you were to mod it to use the 5vsb rail? And regarding the power consumption being greater than the rail can provide, how did ramfan get his to work?

Try not to flame when you answer this :shakes:

EDIT: After a bit of Googling I'm seeing that people COULDN'T get higher voltages by adjusting their rails. I'm not arguing, when I never tried it myself. I'm just looking for answers :)



Sorry, only the first line was meant for you.

The rest of my post was intended for ramfan who told me to "STFU NOOB"


Using the 5vsb instead of 5v CAN be done, but wouldnt be very safe. It would still work and run, but the chances of pushing past what the 5vsb is supposed to put out in terms of whats is pretty high. And I doubt many 5vsb rails have proper over current protection. So it is very possible that ramfan exceeded the power draw of his 5vsb rail.

And then the rails regulation probably started fluctuating wildly outside of atx spec,(for proof read some jonnyguru.com reviews) and damaged his board.

In the end of the day, it very well appears his mod killed his board. Where as my mod using the 5v rail is still kicking fine, and i ran some very high volts through it for benching, and ran with very high 24/7 volts.



PS sorry for not readdressing ramfan instead of you for the rest of my post. Was a bit late and tired.


EDIT: sorry, seen another part of your post. For people to get high vdimm volts by adjusting the 3.3v, they had to first have the 4.0v bios installed. And then they had to adjust the 3.3v rail to be .1-.2v higher than the selected bios voltage. (or you could just select 4.0v and the vdimm would always run .1-.2v below your 3.3v rail. like what I did.

This info was given by Oskar_wu himself at the time. The board's designer.

[[XC] Oj101]

Oh, sorry, I thought the entire post was aimed at me. No worries then :)

[WeStSiDePLaYa]

Well ramfan where are you? I'm looking for your counter argument to what Oskar_Wu has stated himself.

[Cecil]

The 2x256 BH-5s I just got do ~530Mhz 2-2-2-5-7-1T with 3.6vdimm dual channle. That good, average, or bad?

[knopflerbruce]

The 2x256 BH-5s I just got do ~530Mhz 2-2-2-5-7-1T with 3.6vdimm dual channle. That good, average, or bad?

Sorry to say, but it doesn't sound that great. I guess you need better sticks.

[[XC] Oj101]

Bad, that's 265MHz. Mine can do that at 3.1v on a DFI NF3 UT 250GB.

[ramfan]

rebuilding my nf3 250gb, might take a while...
i need a new NB the nf3 ultra chip i got does not work

http://www.livestream.com/redhat_ownage/video?clipId=pla_b900a2c1-8265-47c2-b5e4-15e1977d32f5

EDIT: your still wrong

EDIT2:
im just going to buy this and get it over with
http://cgi.ebay.com/New-NVIDIA-nFORCE-3-nFORCE3-NF3-250GB-Chipset-/250607140478?cmd=ViewItem&pt=LH_DefaultDomain_0&hash=item3a59597e7e

7-15 days shipping....

[Ket]

I got I think it was 23ns DDR500 @ 1-1-1-2 in everest like 3 years ago with my old BH5 I had.

[[XC] Oj101]

I can't touch that with 2-0-0-0, anything below 2-2-2-5 is purely academical on A64. You sure it wasn't 33ns? It sounds more in line with what you got.

[Ket]

I can't remember exactly, too long ago. Those results were with the "new" BH5 which Twinmos got hold of. I loved those BH5s.. So I bought another set that did even better :p Funny thing is my Twinmos sets of BH5 I only paid £50 for and my first kit I ebayed and some dummy paid £150 for the very same kit I paid £50 for lol :D

[Cecil]

Bad, that's 265MHz. Mine can do that at 3.1v on a DFI NF3 UT 250GB.

That wouldnt be dual channle then.

[jmax_oc]

Bad, that's 265MHz. Mine can do that at 3.1v on a DFI NF3 UT 250GB.

Strictly impossible

An average 2x256Mb kit make 260MHz 2-2-2-5 Spi 1M @ 3.5V.
My best stick (not kit) made 260MHz 2-2-2-5 Spi 1M @ 3.3V which is pretty exceptional (more than 80 sticks tested).

[[XC] Oj101]

Correct, so knock off 5MHz and say almost the same speed at half a volt less ;)

[[XC] Oj101]

Strictly impossible

An average 2x256Mb kit make 260MHz 2-2-2-5 Spi 1M @ 3.5V.
My best stick (not kit) made 260MHz 2-2-2-5 Spi 1M @ 3.3V which is pretty exceptional (more than 80 sticks tested).

Anything's possible ;) Here's a PiFast at 262MHz http://hwbot.org/community/submission/908933_oj0_pifast_athlon_64_3000_newcastle_s754_46 .67_sec and even with Trtw=1

That board can only give 3.1v, and that is my best stick ever (I found it last year and I've been doing BH-5 since about 2004 - I dare say I've been through more sticks than you :p: )

[Cecil]

So then 265 isnt bad, its just not great.
My 512 set is pretty good for a 512 set, and these 256s do about 15Mhz more then them at the same volts. With 3.2v I can get 250 dual channle.

[madmaxx]

O/T maybe :shrug:

I had no idea what i was doing but i remember using ocz VX running 250Mhz @ 2/2/2/5 on my p4p800e DLX using a OCZ ram booster to feed 'em 3.7v lol

I eventually fried the ram and one ram slot on my board :rofl: I still have the board up & running and the ram booster is tucked away in the closet somewhere hahaha :up:

OC'ing was SOOOO much easier then, just feed it volt's!

Good time's! :yepp:

[karbonkid]

rebuilding my nf3 250gb, might take a while...
i need a new NB the nf3 ultra chip i got does not work

http://www.livestream.com/redhat_ownage/video?clipId=pla_b900a2c1-8265-47c2-b5e4-15e1977d32f5

EDIT: your still wrong

EDIT2:
im just going to buy this and get it over with
http://cgi.ebay.com/New-NVIDIA-nFORCE-3-nFORCE3-NF3-250GB-Chipset-/250607140478?cmd=ViewItem&pt=LH_DefaultDomain_0&hash=item3a59597e7e

7-15 days shipping....

How are you going to reball it?

[[XC] Oj101]

O/T maybe :shrug:

I had no idea what i was doing but i remember using ocz VX running 250Mhz @ 2/2/2/5 on my p4p800e DLX using a OCZ ram booster to feed 'em 3.7v lol

I eventually fried the ram and one ram slot on my board :rofl: I still have the board up & running and the ram booster is tucked away in the closet somewhere hahaha :up:

OC'ing was SOOOO much easier then, just feed it volt's!

Good time's! :yepp:

Hmmm, yes and no. Remember TCCD? Lowish volts for greater frequencies than BH-5? D9GTS / D9GTR / D9JNL etc where again you just feed it volts (DDR3 at 2.3v and happy :D ).

[ramfan]

How are you going to reball it?

it comes balled all i need to do it line it up and add hot air


Hmmm, yes and no. Remember TCCD? Lowish volts for greater frequencies than BH-5? D9GTS / D9GTR / D9JNL etc where again you just feed it volts (DDR3 at 2.3v and happy :D ).

im running D9GKX right now 2.15V 1000mhz 4-4-4-5 ive had them upto 2.5V 1250mhz same timings

[[XC] Oj101]

What I was showing is old DDR400 happier at lower voltages and new DDR3 happier at higher voltage. Here's one for you:

http://valid.x86-secret.com/show_oc.php?id=417271

[Snowman89]

What I was showing is old DDR400 happier at lower voltages and new DDR3 happier at higher voltage. Here's one for you:

http://valid.x86-secret.com/show_oc.php?id=417271

Well tccd ran higher freq but had crappier timings so BH-5 was still faster.

My bh-5 did 280-285mhz 1.5-2-2-5 with 3.7-3.8v on my Dfi lanparty SLI-D

[ImportantAwareness89]

You guys are all making me nostalgic reading this.... I traded my OCZ VX for OCZ Platinum 2x1gb. I remember 2x512mb was getting a little low and when OCZ realized their VX BH-5 chips had too many issues they discontinued it and started pushing us on other chips. They first offered me DDR-600 chips that had craptastic timings, but I opted for the Plat ones (TCCD Samsungs???) which were rated 3-3-2 or something @ DDR-500. I saw Anandtech bench them at 2.5-3-2 or something so I jumped on the trade up. 1gb => 2gb for free courtesy of OCZ was awesome.

Anyway, I still got TWO DFI NF4 Ultra-Ds in my house. Hell I'm typing on one now. As soon as this i7 system is qualified I'll have a free Ultra-D. I also got some Mushkin Redlines in my mom's system, but I don't believe those are BH-5s. Either way I'll give her my 2gb sticks and play with the Redlines and see what I get :D mahahahahaha



Hmmm, yes and no. Remember TCCD? Lowish volts for greater frequencies than BH-5? D9GTS / D9GTR / D9JNL etc where again you just feed it volts (DDR3 at 2.3v and happy :D ).

I got some Samsung TCCDs I believe. I'm running 2.7 or 2.8 volts on these guys to hit DDR-500. My BH-5s needed 3.2-3.3.



Well tccd ran higher freq but had crappier timings so BH-5 was still faster.

My bh-5 did 280-285mhz 1.5-2-2-5 with 3.7-3.8v on my Dfi lanparty SLI-D

I always wondered about CAS 1.5... never really tried it. I remember DDR500 2-2-2-5 vs DDR600 comparisons at OC Forums with TCCD winning:
http://www.overclockers.com/forums/showthread.php?t=403957

But granted if you can push your BH5 to 270-280 or even higher I'm pretty sure it'll whip TCCD. I was wary of not going above 3.65 volts though (my fingers wanted to type 1.65 as that's what I'm trying to keep my DDR3 under... and I had to do a double take to make sure we did run DDR1 at 3.6volts.. LOL)

[The Nemesis]

Here's what I could find of mine

BH5

http://img259.imageshack.us/img259/4965/bh5screenie0hl.jpg (http://img259.imageshack.us/i/bh5screenie0hl.jpg/)

Uploaded with ImageShack.us (http://imageshack.us)

Samsung TCCD

http://img341.imageshack.us/img341/592/screenie37wg.jpg (http://img341.imageshack.us/i/screenie37wg.jpg/)

Uploaded with ImageShack.us (http://imageshack.us)

[madmaxx]

Hmmm, yes and no. Remember TCCD? Lowish volts for greater frequencies than BH-5? D9GTS / D9GTR / D9JNL etc where again you just feed it volts (DDR3 at 2.3v and happy :D ).

I do indeed recall TCCD, able to achieve almost equal speed as BH5 but with less volt's and lil' slower timing's IIRC

I think that is what OCZ replaced my BH5's with > OCZ5002048EBPE-K :shrug: I also have a set of OCZ5331024ELPE-K in the closet somewhere :rofl: Or maybe they swapped me the 533's and i picked up the 2g 500 kit later? I can't recall exactly as it's been SOOO long. Did the OCZ Vx's come with BH5's on the 2g kit or were they CH5's?

It's been a while and i could be wrong and the chip's on both of these kit's may be TCC5 :shrug:

:banana::banana::banana::banana: i remember cracking 7000 on 3d05 with my 3.0 presshot and a x850xt pe lol The only thing missing from that setup is the BH5's and as a matter of fact that PC is still in use as my backup/bedroom rig! Albeit with a 3850 now

like i said earlier, GOOD TIME's! :up:

[[XC] Oj101]

A little bit of my collection http://www.classicplatforms.com/forum/viewtopic.php?f=73&t=5726

[karbonkid]

it comes balled all i need to do it line it up and add hot air

Awesome. Keep us updated on this please, it'd be nice to see some before/during/after pictures :]

[George_o/c]

A little bit of my collection http://www.classicplatforms.com/forum/viewtopic.php?f=73&t=5726

That's a lot of benching Oj :yepp: :up:

[[XC] Oj101]

Oh, no, not at all, those are just my proudest moments :)

[WeStSiDePLaYa]

rebuilding my nf3 250gb, might take a while...
i need a new NB the nf3 ultra chip i got does not work

http://www.livestream.com/redhat_ownage/video?clipId=pla_b900a2c1-8265-47c2-b5e4-15e1977d32f5

EDIT: your still wrong

EDIT2:
im just going to buy this and get it over with
http://cgi.ebay.com/New-NVIDIA-nFORCE-3-nFORCE3-NF3-250GB-Chipset-/250607140478?cmd=ViewItem&pt=LH_DefaultDomain_0&hash=item3a59597e7e

7-15 days shipping....

So you are going to go and say you know more about this board than the man who designed it.

You truly are an arrogant fool.

You're inability to counter a single on of my points shows your claim has zero basis. Accompanied by you're choice to continue denying even though you have no evidence, shows you aren't a big enough person to admit when you are wrong.

Noob.

[TiN_]

:D I found this thread.

That was 5 years ago, omg.

I've soldered own 512MB sticks from few 256MB BH-5 mushkins, wanted to have 2 slot populate only with 1GB of 250MHz+ BH-5's runnin at 1T.
There was no way otherwise to have 512MB DR sticks

So I've took my 256MB crap and soldered off chips and put them on second PCB. Then reflashed SPD with new bank configuraion settings and viola, gotha 512MB stick able to run 250MHz+ with only 3.5V.

Used DFI LanParty NF4 Ultra modded to SLI that times. Maybe some folks like njkid32 remember me here still :D

[ramfan]

:D I found this thread.

That was 5 years ago, omg.

I've soldered own 512MB sticks from few 256MB BH-5 mushkins, wanted to have 2 slot populate only with 1GB of 250MHz+ BH-5's runnin at 1T.
There was no way otherwise to have 512MB DR sticks

So I've took my 256MB crap and soldered off chips and put them on second PCB. Then reflashed SPD with new bank configuraion settings and viola, gotha 512MB stick able to run 250MHz+ with only 3.5V.

Used DFI LanParty NF4 Ultra modded to SLI that times. Maybe some folks like njkid32 remember me here still :D

i remember this :P


So you are going to go and say you know more about this board than the man who designed it.

You truly are an arrogant fool.

You're inability to counter a single on of my points shows your claim has zero basis. Accompanied by you're choice to continue denying even though you have no evidence, shows you aren't a big enough person to admit when you are wrong.

Noob.

ok so you want me to show you 4v at the vdimm mosfet with only 1 resister mod?

the way oskar designed the vdimm supply was to allow it to be modded very easy to allow greater than 3.1v
the only reason the nf3 was not sold with ability to run high vdimm was because of heat issues with the mosfets so they limited the the input voltage to 3.4V with 2 resisters.

[Kosmic71]

I do indeed recall TCCD, able to achieve almost equal speed as BH5 but with less volt's and lil' slower timing's IIRC

I think that is what OCZ replaced my BH5's with > OCZ5002048EBPE-K :shrug: I also have a set of OCZ5331024ELPE-K in the closet somewhere :rofl: Or maybe they swapped me the 533's and i picked up the 2g 500 kit later? I can't recall exactly as it's been SOOO long. Did the OCZ Vx's come with BH5's on the 2g kit or were they CH5's?

It's been a while and i could be wrong and the chip's on both of these kit's may be TCC5 :shrug:

:banana::banana::banana::banana: i remember cracking 7000 on 3d05 with my 3.0 presshot and a x850xt pe lol The only thing missing from that setup is the BH5's and as a matter of fact that PC is still in use as my backup/bedroom rig! Albeit with a 3850 now

like i said earlier, GOOD TIME's! :up:

Good times indeed! OCZ Plat rev2 2x512 (I still have the sticks!) :D

http://img541.imageshack.us/img541/3376/fil9552.jpg

[knopflerbruce]

I got some 2x512mb PC5000 DFI Edition sticks in the mail the other day:)

[[XC] Oj101]

Oh? Sounds like OCZ DDR625 to me :D

[knopflerbruce]

Oh? Sounds like OCZ DDR625 to me :D

Indeed it is:p: Paid 20 bucks shipped:D

What do you think of my avatar, btw?

PS: stop spamming XS and get your ass in the chatbox a little more often!!!:D

[[XC] Oj101]

Holy $%^@ what happened to your avatar?

Hahaha, I typed that before even reading your post. Yessir, on my way...

[MaSell]

I've played my toys this weekend:

http://img36.imageshack.us/img36/2646/beha5.png
CPU-Z Validator 3.1 (http://valid.canardpc.com/show_oc.php?id=1346670)
4v

http://img839.imageshack.us/img839/8056/bh5.png
http://valid.canardpc.com/show_oc.php?id=1346736
Just 3.85v

http://img541.imageshack.us/img541/1089/512p.png
As little as 3.65v

http://img576.imageshack.us/img576/1851/98684273.jpg

http://img709.imageshack.us/img709/852/62773489.jpg

Nothing unusuall, but still interesting, especially in 2010 :)

[M.Beier]

Crap SellMa... Seems I have to get a nF4 Ultra-D and a 3700+...... I have to beat you...
286mhz, damnit.....