but thats only with 512k cache which means that they only need 2.4G to compete in 3D
Printable View
but thats only with 512k cache which means that they only need 2.4G to compete in 3D
All three were pretty much exactly the same as far as I could tell. I am begining to think that the multi is KING in A64. I might be joining you guys on friday.
first off rabbi. i wish i could give you a good answer. i never fully tested my tccds sticks to their max with my 3400. i just played around with them till i found a good spot.(sign of a noob overclocker) i will see if i can get some decent low multiplier high 1:1 fsb notes here by saturday or sunday.
updates:
-been working for 2.7 at between 1.58-1.65 but can't seem to get it. (my original hopes were 2.7 at 1.5v)
-right now i am super pi 32m stable at 1.55x1.04 @ 294x9=2646. (ram at 2.5-3-3-10 1T 264.6 Mhz) i am testing prime95 small fft and am going to let this go for a while.
benchies@ this clock:
superpi 32M: 29m 18s
3dmark 2001 with x800xt at stock 500/500: 29373 (always a nice score with the 1mb cache)
Sisoft Sandra:
Arithmetic: 12241 Dhrystone, Whetstone 4181/5419
Multimedia: 25245 Integer, Floating point 27149
CPU is staying in the mid-upper 30s (c) under load. Room temp is around 25 c.
i have alot of stuff i have to get done for school tomorrow but i will try and update if i get any furthur. ff, i hope you start getting better numbers. i wouldn't be very happy at 2.3-2.4 with 1.5-1.65v.
I think I'm starting to get somewhere. After doing some tweaking, changing Vdimm from 2.6v to 2.7v seemed to help me break the hump that was holding me back. Right now I'm Priming at 2.4GHz with 1.5v (didn't seem to like this speed with 1.4v). Will keep pushing it.
Have you guys tried running with a divider just to find max CPU speed?
Yeah that's what I'm doing. I set the board at 300HTT and divided everything to make sure the board was stable. Keeping the RAM at or below 200MHz for now, and slowly working the CPU multi up in 0.5x increments (hopefully) to 2.7GHz.
1/2 multi's are evil though :(
just because i can't seem to figure it out, tristian and FF, you both have MT-34's, right?
i don't really understand the discrephancy in oc'ing.. though i guess the comment before that the CTO made about slower transistors.. is proving right :mad:
I dunno, I think the chips are about the same, I just hadn't really tweaked it enough to keep up with tristancarton. I'm currently at about 2.6GHz now, but don't see this chip going much further. Am trying to find out how high I can run the memory, as I've never really pushed it beyond 250MHz.
mats, i started out doing that but i wasn't really getting any furthur running the memory at low speeds. (running the divider at 100 (1/2) or 133 (2/3)wasn't giving me a whole lot more room than running it at 166 (5/6) or 180 (9/10).
ozzi, yeah strange to see a few san diego reviews showing very high clock speeds at low volts and we are sitting here trying to get 2500-2600 stability in the 1.5v-1.6v range. aren't these lancaster turion cores basically voltage binned san diegos that are cut down to only use the single channel memory interface? i still have a hard time believing amd would use special silicon for these chips when alot of their newcastle 90nm cores can hit the turion low voltage speeds. then again with the results we are getting so far.
yeah ff has an mt-34. supposedly of a similar batch.(makes sense since they came from same person and one shipment) lastly give us a few more days to play around with the cpu some more. perhaps there are a few memory timings and settings that just don't get along with these new memory controllers and ff and I haven't been tinkering with those yet. (right now i am just more or less just running oskar nf3 tccd settings.)
small update: keep having to move my speed down a few from prime95 failure. testing 292x9 just failed after a few hours. heading for 290x9.
fr0stedflakes,
can you post your complete settings mate?
maybe we can help you get the board stable so it's not a limiting factor...
For now:
Set LDT to 3x
Set CPU Multi to MAX ( 9x ??? )
Set Memory Divider to 166 (5/6)
What memory do you have? We will definately be able to help you get that under control if you post what manufacturer, what chips, what speed ratings, yada yada.
What PSU do you have?
what BIOS are you running?
If you need a hand just post that info and we'll be Xtremely happy to help hehehe
rabbi i may end up posting alot of questions later tomorrow and asking for thoughts. i know you have tweaked your nf3 to the bone.
anyway, more comments.
-290x9 (2610) was stable at 1.55x1.04 with 2.5-3-3-10 1T cpu/10 (261 mhz) timings for 5 hours in prime small fft. (i stopped it)
-296x9 (2664) at 1.5x1.10 with 2.5-3-3-10 1T cpu/10 (266.4) went for about 2 1/2 hours (p95 small fft) last night before it exited out.
-right now trying 294x9 at 1.5x1.10 with 2.5-3-3-10 1T cpu/10 (266.4).
-these chips are really interesting. i was throwing 1.7 volts at it earlier and the temperature would only max at 38-40c. additionally right after taking off the load the idle cpu would drop back down into high 20s. i thought this was pretty amazing for air cooling temps at 1.7v in a room with ambient being in mid 20s. highly doubt i will run 1.7v even with those temps. 1.6-1.65 is probably even pushing a 90nm air cooled core.
ff how are your temps?
Temps are awesome. My Oakville wasn't much different. Even with pretty low end air cooling, you could probably give these 90nm cores 1.55-1.6v and still keep temps under 50*C load. But like you mentioned, I'd be worried about the high voltage stressing the processor, and not the heat. I'd say 1.6v is an OK 24/7 Vcore, maybe 1.55v if you want to play it safe.
I doubt it's the board or mem, because right off the bat I set it to 300x7 to verify stability at high HTTs. Divided the memory to ~200MHz (2/3), and set LTD to 2.5x. I tested this and it was stable in P95 and general windows use. Because my goal was 2.7GHz I was just planning to keep all settings the same and raise multi so I'd know it was the CPU as soon as I started having instability. Mem frequency would change a little as I started OC'g the CPU because it's divided by the CPU frequency, but it never went over 200MHz, which I knew for sure it could run.
Tried BIOS 4/8 and 4/15. My power supply is a 350w, but it's an Antec with plenty of power. Is rated for 21A on the +12v and my DMM reads +12v as 12.01, so I doubt that's the problem. I have PCI video, 2x512mb, and CD-RW and a 40gb HDD, so I don't have a lot of big power users in my system.
I've taken a break from tweaking for a few days because it's getting me so frustrated. I'll give the processor a couple days to burn in then go at it again. Or maybe I'll just drive over it with my car... :hehe:
Quite disappoing, isn't it ? :confused:
Now i started to glorify my NC 2800+ once again :woot:
BTW: is the multiplier realy unlocked or I've overheard it ?
That is a very weak PSU for nF3 bro, it would have been "acceptable" on nF2 but I wouldn't go running it on your rig and expect great overclockingQuote:
Originally Posted by fr0stedflakes
Hey Guys!
I am really sorry that these things turned out to be Turdions. :(
I was able to talk my distributor into sending me an MT-34 instead of an ML-37, so I am in this with you guys now. I tried the Turdion without flashing the K8V with a new BIOS and guess what? It worked! Check out my setup! LOL
http://img9.echo.cx/img9/9589/k8v2yo.jpg
http://img9.echo.cx/img9/2117/k8v20ji.jpg
A nice and cool 35C in a room at 25C.
Wow, what heatsink is that!?? Is it better than a XP90?Quote:
Originally Posted by Vandi423
Wow man, you NEED a better PSU for OCing. Also, use 3x LDT multi instead of 2.5x. 3x is more stable.
Quote:
Originally Posted by fr0stedflakes
lol, thats the only ausus board that actually looks good.
but the pci/agp isnt locked is it?
and i see you have used pads on the pcb. are those from an AXP?
Well the K8v's that I have were free. So no OCing on these and yes they're from a palomino.Quote:
Originally Posted by reject
link for the 4/15 bios update plz???
hopefully there is hope for the ML's, as the MT's dont seem to be breath taking
vandi, no need to be sorry, we knew what were getting into.(although that doesn't mean that we aren't surprised at the results) nice to see you joining ff and i with an mt-34. these things run really cool. would be perfect for a htpc or as a upgrade for a laptop running a 81w dtr or 62w mobile.
i found nemo, here is the 4/15 dfi bios
westside i bet the mls are on the same process as the mts. i originally figured they were using the san diego process but the latest reviews of the san diego chips combined (1.4-1.5v 3ghz water) with our not so amazing results may indicate amd has it's own process/si wafers that are used for these turion chips.(what the cto said but i didn't believe him)
another good question that we need to get from an amd insider is whether this latest mobile 3700 nm chip will be based on the turion (lancaster process) or the san diego process. (if there is a difference between the two processes)
last, update on speeds. system was prime95 small fft 12+ hour(i stopped it) stable at 2646 1.65v. however would not pass 3dmark 2003 at that speed.(i am quite confident that my antec 550 is being squeezed for all it's worth. prob be picking up a new ps soon.) computer will pass 3dmark 2003 at 2600 1.65v. i think i may loop one of the graphic programs overnight to test stability.
350w may not be adequate for P4 Prescott and a 6800 Ultra, but I seriously doubt that my PCI video card and 25w processor is pushing it to it's limits...
Even overclocked, I doubt this Turion is drawing more than 60-70w.
wow that bios update helped me out with my ram a lil bit, go figure thanx tris.....anyways guys, have you tried a freash rebuild now that the cpu is burnt in a little take the whole thing apart put it back together and re-do windows freash install and partition, also tried with ide drives?????? and try it with one stick of ram.....the dfi's o/c better with 1 stick and also if you are using sata try switching it around, i think you can't hook it upto 1&2 .... dunno hope this helps
So....whats the final verdict on these ??? Are they Dothan Stompers ???
EDIT: Oh hey FoundNemo...
Perkam