3800+ Venice Screenies (Post #1 = FAQ)

[MrQ3W]

My Newcastle has survived 2.16V pumping through it for an hour with only the stock cooler to protect it.

[DrJay]

In the first post I wanted, but forgot to say "He (Geforce4ti4200) probably meant increased performance and not actually higher IPC".

Makes sense. Higher performance but most likely not higher IPC.

[Tekdemon]

dammit I was all set to buy a Winchester 3000+ but now I'm tempted to wait for Venice to come out in 3000+ trim...
Although I guess we don't really know how the yields are going to be yet.

[krille]

What's up with the benches? We're waiting

No, really, take your time... and blast the thing to HELL!!

[trakslacker]

yeah, I'm almost certain his screenie was just to show that Venice has no subzero boot issues. give the man time, I'm sure he'll give us what we want to see.

And as much as I hate to say this, stephen, go get some sleep!

[clockedOut]

dammit I was all set to buy a Winchester 3000+ but now I'm tempted to wait for Venice to come out in 3000+ trim...
Although I guess we don't really know how the yields are going to be yet.

I really dont see why anybody wants to buy a winchester??? The newer ones seem to be poor clockers and have weak mem controllers.
UNLESS you can find the MAGIC cpu.

[terrace215]

Right. Very little because they were optimizations, which is what you are speaking of. I'm not an expert in this field but to the best of my knowledge, to change the instructions per cycle, you'd need to alter the architecture of the cpu.

No, you certainly aren't.

Improved performance at the same clock = more instructions per cycle. The global architecture does not need to be changed. Optimizations "count" too.

[TommyW]

than can we buy this amaizing cpu

i think i wait and only buy a new mobo , video card , memory , case

and after that i buy me new vanice 3200+

[Kunaak]

amd may do tweaks to the cache, ondie controller and what not. I already saw benchmarks where a winny was 3-5% faster than a newcastle(both s939) and I hear venice will once again raise this

the reason why this happend is not cause of anything being tweaked or really changed in any signifigant way.
it's cause they moved from the .13 process to the .09.
thats all.

[TommyW]

less heat more pref...

[shuRe]

i agree with Ailluer. So it can do 3g underphase which is OK. It does 2.9 on air stable? (maybe) which is about a 2.7g fx55.

fx55 will do 2.7 on air easy, prolly 2.8-2.9 if ur lucky, and seeing that this is a 3800+ i cant see it being that cheaper than a 130nm fx55..

If the 3500+ could do 3g stable on air then i would be impressed.

[Caper]

look at this
already avaible?

Nope, just because it´s listed doesn´t mean it´s available.
They have none at stock.
A company here in Sweden has also listed it and expect it to be available April 26.

[s7e9h3n]

Okay ill think ill be getting rocks thrown at me but i really dotn see what the big deal is.
First, its already a 3800+, a 450mhz overclock (and thats only superpi 1mb stable) isnt anything to write home about. Hell if it stops at 2850mhz (he said it didnt finish a 2900, so thats less than 30 seconds of stability), a good 3000+ can already do that.

Then 3000mhz at -47? Sure he didnt throw 1.70volts at it but im not sure these will even TAKE 1.7v, theres always a threshold where upping the voltage actually does anything, and if the stock voltage is lower, then the upper voltage will follow.

I guess im gonna have to see what a 3000 or a 3200 can do cause a 18% overclock (that we dont know is stable) on air isnt really blowing my pants away.

Hey, if you cared to read the thread, you'd understand that the "big deal" doesn't concern pure clock speed as opposed to basic characteristics of the new Venice core. Show me your Winchester that can boot at -30C and we'll talk, otherwise just continue to read and you may get a better understanding of why this thread is here in the first place. Btw, I DID throw 1.7V+ at the cpu - I just haven't posted the results.....yet....

[DrJay]

No, you certainly aren't.

No need for that. We can agree to disagree,.......at least until one of us is proven wrong.

Improved performance at the same clock = more instructions per cycle. The global architecture does not need to be changed. Optimizations "count" too.

I think improved performace can be achieved by things like more efficient handling / routing of data. It seems like such things could result in increased performace to you, the user of the computer without actually increasing the internal IPC of the cpu.

[Caper]

Btw, I DID throw 1.7V+ at the cpu - I just haven't posted the results.....yet....

What are ya waiting for??

[andL64]

yeah but none in stock its also listend @ geizhals 3x 3800+ 1x .13 2x .09 cores one is a venice with BP stepping not really available.
Still one shop has it listend in .de they also got listend a 4000+ "winchester" but with 1024kB? maybe/4sure? its an error because the shop listend only one 3800+ which is a "winchester".

gonna know moa bout in some days also whats about 4200+ and so on

pretty nice times for AMD it has already done cores that finally really really rocks!

seEn

[RAndomaN]

getting better, show us what this thing can really do stephen :p

[s7e9h3n]

What are ya waiting for??

Hold up real quick - you ever had one of those long OC'ing nights where you wake up the next day and everything wasn't as clear as you had thought it was the night before? Well if you think that sounds stupid, then you'll really like the fact that I can't find where I saved some screenies I was sure that I'd taken Excuse me while I search a little bit, but here's a little something to look at. If you guys don't hear from me for an hour or two, it's because I'm going to work and I find it kind of tough to post from my laptop while driving 80+ Mph...please be patient.



Oh, BTW: this is at -29C Idle, -15C Load. Measurements were taken from a thermocouple I attached to the side of the IHS of the 3800+.

[terrace215]

I think improved performace can be achieved by things like more efficient handling / routing of data. It seems like such things could result in increased performace to you, the user of the computer without actually increasing the internal IPC of the cpu.

It's very simple. If you replace the CPU with a different CPU that runs at the same frequency, and the system performance improves ( = faster = same instructions were processed in less time), the new CPU has better IPC than the old one, by definition.

And that's just what the mentioned optimizations accomplish.

[alexio]

Nice results !!!

You can atleast tell us if it likes high volts right?

[Caper]

If you guys don't hear from me for an hour or two, it's because I'm going to work and I find it kind of tough to post from my laptop while driving 80+ Mph...please be patient.

Drive carefully, no rush.
And ya better get some sleep tonight, don´t spend all night at the rig
We are all curious about the Venice but do not mean to rush you.
Take your time and post when you feel to

[stealth17]

looks like there will be a 3000 venice

ADA3000ASBOX 1800MHz 512KB 1.40V 90nm 67W 70° C Venice 0.31
ADA3200ASBOX 2000MHz 512KB 1.40V 90nm 67W 70° C Venice 0.31
ADA3500ASBOX 2200MHz 512KB 1.40V 90nm 67W 70° C Venice 0.31
ADA3800ASBOX 2400MHz 512KB 1.40V 90nm 89W 70° C Venice 0.31

[MrQ3W]

looks like there will be a 3000 venice

ADA3000ASBOX 1800MHz 512KB 1.40V 90nm 67W 70° C Venice 0.31
ADA3200ASBOX 2000MHz 512KB 1.40V 90nm 67W 70° C Venice 0.31
ADA3500ASBOX 2200MHz 512KB 1.40V 90nm 67W 70° C Venice 0.31
ADA3800ASBOX 2400MHz 512KB 1.40V 90nm 89W 70° C Venice 0.31

Sweet.

[s7e9h3n]

Nice results !!!

You can atleast tell us if it likes high volts right?

Yeah, it can handle a decent amount of Vcore, but performance scales oddly in terms of vcore with this chip. I'll explain later when I get to work, but check the v-core here And YES, I did try higher....



I don't recall the exact temps for this shot, but I'm thinking ~ -25C idle, ~ -12C load

[stealth17]

i cant wait to see one of these on a DFI nF4...it will pwn