Athlon64 X2 3800+ Steppings

Big.Wayne · 15 Aug 2007 at 23:05

Hi,

anyone know the difference between:

LCB9E 0632DPMW

&

LCBQE 0650SPMW

I am in the process of overclocking them both but I'm slow and wondered which is likely to overclock the most?

Thanks

[edit] Both from overclockers.co.uk

str · 15 Aug 2007 at 23:08

LCB9E is likely to max out at around 2.7 to 2.8 GHz with stock volts and possibly upto 3GHz with <=1.5v.

LCBQE might do a bit better.

Let us know how you get on.

GeForce · 15 Aug 2007 at 23:13

I'm also putting my money on the LCBQE going further. But I'll say the max you'll get out of either of them is 2.8Ghz, 2.9 is you're very lucky.

I reckon 3Ghz is a tad optimistic. But who knows

We shall see!

Jon

Big.Wayne · 15 Aug 2007 at 23:18

str said:
LCB9E is likely to max out at around 2.7 to 2.8 GHz with stock volts and possibly upto 3GHz with <=1.5v

LCB9E 0632DPMW - 1.375vCore - Rock Solid!

Ah that seems about right, this chip runs 2600MHz at 1.300vCore but then needs 1.375vCore to reach 2700MHz, thinking maybe it may reach 2800MHz but 3GHz seems a bit far off. .

str said:
LCBQE might do a bit better

Good, I thought they may be a dodgy stepping!

Tee Hee Johnson · 15 Aug 2007 at 23:21

i wish i knew what mines was... cant be bothered taking it apart to find out lol

Tee Hee Johnson · 15 Aug 2007 at 23:24

its a 9E...

Found a pic lol

str · 15 Aug 2007 at 23:43

When I was closing in on the max overclock I often found the small ffts test would run fine for hours but the blend would show up problems quite quickly.

Big.Wayne · 16 Aug 2007 at 00:04

Blend is a lot easier on the processor than Small FFTs I find.

Perhaps you were running into memory or chipset issues?

str · 16 Aug 2007 at 01:03

I found the small ffts test useful for short runs to find the maximum overclocking speed before it was stopping with an error.

The problem with small ffts is that it's very forgiving of memory or motherboard limits so even with 24 hours of small ffts passed it could still crash after running a game for 5 mins whereas passing a blend test is much more likely to result in a system that is stable in everything.

Also I usually suggest to use priority 9 as it's sufficient for Orthos to max out the stressing even when the system is being used for other things but priority 1 is fine too as long as the system is only running Orthos (and not used for anything else) and the screensaver etc is disabled.

Big.Wayne · 16 Aug 2007 at 01:17

Yup I know that

But if you got a test rig where the mobo can run over 300MHz-FSB and the memory can run 260MHz (tested and tested again!) and your concentrating on just overclocking the chip, wouldn't you run just the Small FFTs test? (its a tougher CPU test than Blend).

I also have been involved in the priority 1 vs 9 debate (going back nearly two years!) and I will not get involved in that again, suffice is to say the machines are left alone testing, no screensavers, nothing to snatch 'priority' away from Orthos etc, just kinda nice when u have to nip in there for 2 mins that you can actually use the machine for a quick screengrab etc.

I use Large FFTs test when Im just testing memory, I find thats good for finding errors in the memory and chipset that Memtest86+ doesn't.

Only really use 24 hour Blend as 'health check' on my PC's once or twice a year.

Amonlym · 16 Aug 2007 at 10:02

Ive found LCB9E's seem to have a hunger to low temps and high volts..

shaving a few c's off has time and again alowed me to drop v-core. 2920 @ 1.4v as long as temps below 45c load. need 1.475 above 55c load!

str · 18 Aug 2007 at 21:53

Big.Wayne said:
Yup I know that

But if you got a test rig where the mobo can run over 300MHz-FSB and the memory can run 260MHz (tested and tested again!) and your concentrating on just overclocking the chip, wouldn't you run just the Small FFTs test? (its a tougher CPU test than Blend).

I also have been involved in the priority 1 vs 9 debate (going back nearly two years!) and I will not get involved in that again, suffice is to say the machines are left alone testing, no screensavers, nothing to snatch 'priority' away from Orthos etc, just kinda nice when u have to nip in there for 2 mins that you can actually use the machine for a quick screengrab etc.

I use Large FFTs test when Im just testing memory, I find thats good for finding errors in the memory and chipset that Memtest86+ doesn't.

Only really use 24 hour Blend as 'health check' on my PC's once or twice a year.

I was surprised at the 24 hour small ffts test but now you've explained the conditions of your testing I guess it's not such a bad idea. Also I hadn't even considered the large-ffts test over the blend so I'll probably give that a try next time I'm testing.

Big.Wayne · 18 Aug 2007 at 22:16

str said:
blend

I believe this test is a 'blend' of the Small FFTs and Large FFTs, so it gives your CPU a good hammering then changes to giving your memory a good hammering but I don't think it does both together?

I say this because when I used to run the Blend test I noticed that my CPU load temps would drop quite a bit during testing whereas the CPu temps are consistant during Small FFTs (albeit with a little fluctuation due to changing ambient conditions).

Anyhow I'm still testing lol, pleased to say both the LCBQE steppings have passed a 24 hour run of Small FFTs @2700MHz using 1.3000vCore (the LCB9E chip needed 1.3750vCore to run stably at the same speed!).

now working on 2800MHz but the chips are starting to get thirsty, up to 1.350vCore on one chip and 1.3750vCore on the second (both LCBQE), although to be fair I have observed on both platforms that the chips are both reading 1.3750vCore under load in Windows so theres a little variation in voltage on both machines. . .

Will post back some more results soon . .

str · 19 Aug 2007 at 00:01

Excellent results so far and temps etc look good so should go even further.

Big.Wayne · 22 Aug 2007 at 00:32

Slowly getting there!

All the chips will do 2600MHz at 1.3000vCore

The LCB9E is up to 2700MHz @ 1.3750vCore

The LCBQE is up to 2700MHz @ 1.3000vCore

One LCDQE is up to 2800MHz @ 1.3625vCore (pictured above, using stock air cooling).

GeForce · 22 Aug 2007 at 00:56

Big.Wayne said:
<snip>

That's impressive! Nice work :cool:

bigjonnyauk · 29 Aug 2007 at 18:50

wayne did you buy both of these due to different steppings to try reason i ask is that i mailed overclockers to see if the batch they have in are e6 as thats what i need and was mailed back saying we dont know see what it is when u get it lol thats no good to me

so if you are getting rid of the one that dosent clock great then gimme a shout

BigglesPiP · 29 Aug 2007 at 18:54

Is one of those the 35W SFF special?

Mul · 30 Aug 2007 at 11:07

Unlikely, AMD never produced 35W 939 X2's. If I recall correctly the original X2's were 115W models at 1.4V, while some 65W/89W (1.3/1.35V) models appeared later on. Big.Wayne's is likely to be one of the latter, probably the 1.35V version.

GeForce · 30 Aug 2007 at 11:57

The original X2s (all E4 Manchesters/Toledos/Denmarks) are "ADA" and have 110W TDP and 1.4v stock Vcore.

The newer X2s (all E6 Manchesters, some Toledos/Denmarks) are "ADV" and are 89W versions with 1.35v stock Vcore.

The 65W versions are the "EE" editions and IIRC, are only available on socket AM2.

Jon