-
We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.
Athlon 64 3000 + 939 90nm AMDZone Review!
- Thread starter AmdEmAll
- Start date
charloscarlies
Golden Member
I've been telling people this for months now. 🙂
Although, I'm still looking forward to how well these 90 nm chips overclock.
Zebo,
I noticed the 3000+ in your sig...is that on air? Newcastle I would assume? Bet it screams...nice overclock. 😀
Well, this is just a shrink from .130 to .09, no other changes. There were rumors of added SSE3 and increased performance, but if that happens it will be in a later revision. I don't blame them either. Make one major change at a time to avoid trouble shooting IF a problem occurs. Once they get .09 worked out, then they can modify the core.
My .02 🙂
My .02 🙂
LifeStealer
Senior member
That looks VERY impressive compared to the other chips and price. With stock cooling thats a pretty good overclock too, can't wait for the next revisions. Looks like AMD learned a bit by watching Intel and IBM.
charloscarlies
Golden Member
Am I the only one that isn't very impressed by those benches? It seems about on par with a 754 2800+...yet costs $40-50 more. That overclock of theirs was pretty weak...stock cooling or not. Obviously I'll wait for more people to get their hands on these...but color me unimpressed.
This just goes to prove my point that waiting for the performance "increase" of 939 and 90 nm was a waste. Definitely glad I went with 754 months ago. 😀
This just goes to prove my point that waiting for the performance "increase" of 939 and 90 nm was a waste. Definitely glad I went with 754 months ago. 😀
Zebo
Elite Member
They overclock probably the same due to SOI http://home.comcast.net/~jiahwa/. 90nm 939 vs 130nm 939 are about the same performance.Link Sad to say with the slight edge going to "old" 130nm. Certainly did'nt live up to it's 'core imporvement' hype found in plenty of threads around the net. That's all I'm saying.
Another thing is AMD is doing a terrible job of labeling these 939's that are 200mhz slower than the 754 with the same PR ranking. Ie 3000 vs 3000 1.8 vs 2.0 as seen in the AMDzone review where 754 annihilates the 939 3000. Should be more like 2850 for the 1.8Ghz 939 not 3000.
People need to look at 90nm for what it is. Cost savings and making money not some revolutionary new technology. When AMD or intel can put 35% more chips on a $1500 waffer it increases thier bottom line not nessesary beneficail to John Q Public (P4E) due to gate leakage at these sizes. AMD has chosen a more expensive route called SOI to minimize this problem and by initial accounts is sucessful the 90 and 130 run about the same temps.
Anyway, the AX skt 754 newcastles, which I have, have been reaching 2.5Ghz w the stock HSF for several months now, only a blind man could'nt see it with a cursory review of various forums. Fine 939 is faster, but you'd have to get one clocked @ 2400-2425 mhz to equal the 754 OC due to only 3-5% advantage for it dual channel. Not worth it at over double at the price for several months now. However today it is worth it if monarchs price is accurate.🙂
Water but air equivalent really. I don't have any fans to speak of on my radiator so it really does'nt count as water perse. Probably just fractionally better than the best air solution.
charloscarlies
Golden Member
Good to hear. I was all excited when I ordered my 3200+ NC from Monarch only to find they had sent a CH instead. It is a CG...but it won't do over 2.3 ghz on any voltage. This is with a Zalman 7000 so cooling isn't "bad" really.
I'm putting together another rig...and I'm pretty set on another 3200+, but making sure I get a NC this time. In your opinion...do you think the 3200+ is really worth the extra money? It seems that most of the Newcastles are hitting around 2.4-2.5 ghz no matter what the PR rating is. Although...the idea of the 3400+ clocked stock at 2.4 is kind of enticing. What would you get if you were in my shoes? (price is an issue 😀)
Zebo
Elite Member
A3200 clawhammer (ADA3200AEP5AP)? You sure?
All 3200 nc are true nc (ADA3200AEP4AX) so they obviousy sent you the wrong chip!! RMA that bad boy to Joe. Unless you specifically ordered a clawhammer they sent you the wrong chip.
There is some confusion in the 2800 and 3000's where they used old cut, disabling 512 cache, crappy CG clawhammers to make them. But thier is absolutly no confusion in the 3200 and 3400's. They are either a clawhammer or newcastle.
Make sense? Probably not.
Anyway if your putting together another rig as a nessesity I may wait wait untill more retailers get these low clocked 939's and see how the price hashes out. I mean If you can get a 3000 939 for $180, and it's been as low as $190 at Joes site, then that's the better deal. Otherwise I would skip on both 3200 and 3400 NC if your overclcoking diue to price premium and get: ADA3000AEP4AX form $170 from newegg...after all they are the same chips who cares what AMD labels them as and charges for them 😀
All 3200 nc are true nc (ADA3200AEP4AX) so they obviousy sent you the wrong chip!! RMA that bad boy to Joe. Unless you specifically ordered a clawhammer they sent you the wrong chip.
There is some confusion in the 2800 and 3000's where they used old cut, disabling 512 cache, crappy CG clawhammers to make them. But thier is absolutly no confusion in the 3200 and 3400's. They are either a clawhammer or newcastle.
Make sense? Probably not.
Anyway if your putting together another rig as a nessesity I may wait wait untill more retailers get these low clocked 939's and see how the price hashes out. I mean If you can get a 3000 939 for $180, and it's been as low as $190 at Joes site, then that's the better deal. Otherwise I would skip on both 3200 and 3400 NC if your overclcoking diue to price premium and get: ADA3000AEP4AX form $170 from newegg...after all they are the same chips who cares what AMD labels them as and charges for them 😀
charloscarlies
Golden Member
Yeah it was the wrong chip. I called em up and they said they'd do the RMA...but once I realized it was a CG Clawhammer I decided to try it out. It's actually in my brother's rig for now...so I just didn't worry about it. 🙂
I had to RMA my brother's Chaintech VNF3-250 after he fried the AGP slot by trying to put his brand spankin' new 6800 GT in while it was running. Needless to say his rig is WAY overkill for him...but hey he has the money so why not right?
I was planning on using the Chaintech and just picking up a 3000/3200+. ..but these 90 nm's caught my interest. I think I am going to just wait it out and see how these new chips affect pricing on the 754's.
Good to hear that there really isn't a different between the 3000+ and 3200/3400+ though!
I had to RMA my brother's Chaintech VNF3-250 after he fried the AGP slot by trying to put his brand spankin' new 6800 GT in while it was running. Needless to say his rig is WAY overkill for him...but hey he has the money so why not right?
I was planning on using the Chaintech and just picking up a 3000/3200+. ..but these 90 nm's caught my interest. I think I am going to just wait it out and see how these new chips affect pricing on the 754's.
Good to hear that there really isn't a different between the 3000+ and 3200/3400+ though!
charloscarlies
Golden Member
Wow...yet another idea to think about! 😀
The 9x multi might be a little tough to deal with...but this Geil does 274 HTT 1:1 on this Chaintech so the low multi should be ok. The 10x multi of the 3000+ would be ideal though...makes nice round numbers. 🙂
Zebo
Elite Member
Wow what mem is that, exactly? and what timmings?
Still 274 x 9 only gives you a max capability of 2466mhz if you still want to retain 1:1 mem ratios. What if you get a 2600Mhz capable chip like one did in that thread? Then you have to run mem 166 which means 290HTT x .83 = mem running 240..
The 3000 just gives more flexability that's all. It can "look" like a 2800 by still using a 9 multi if nessesary and optimal, BUT a 2800 can never "look" like a 3000 since it's incapable of using a 10X.
Still 274 x 9 only gives you a max capability of 2466mhz if you still want to retain 1:1 mem ratios. What if you get a 2600Mhz capable chip like one did in that thread? Then you have to run mem 166 which means 290HTT x .83 = mem running 240..
The 3000 just gives more flexability that's all. It can "look" like a 2800 by still using a 9 multi if nessesary and optimal, BUT a 2800 can never "look" like a 3000 since it's incapable of using a 10X.
charloscarlies
Golden Member
Yeah I'll prob just stick with the 3000+...although using a mem divider hasn't seemed to hurt performance much from my testing. Still 1:1 looks so purty!
I'm using this Geil Ultra PC4000. It takes the full 2.9 vdimm on the Chaintech...but it does 274 @ 3-4-4-8. Passes prime blend test and several memtest loops. It was only like $240 for a gig a few months ago when ram prices were insane.
Concillian
Diamond Member
If it's in a DFI LANParty UT you set the mem to 180 (9:10) which puts it at 260 MHz, same as if you had a 10x multiplier. 😛 The LANParty really removes most advantage of the 3000+ over the 2800+ with the 9:10 memory ratio. You want to run at the speeds available to someone with a 10x multiplier? Set the memory to 9:10... LoL.
I don't get why the BIOSes don't just let you set the integer divisor? I mean the A64 only uses integer divisors of CPU speed for memory speed anyway, so why not let the end user specifiy the divisor instead of making us figure out what the memory is really running?
Concillian
Diamond Member
What is the default voltage on the 90nm 939s? I was nder the impression that default voltage was 1.4 as opposed to the 1.5v that the 130nm cores have. This would put it as a worse overclock, as Newcastles don't usually need + 0.25v to get 2500 MHz.
I knew I had read that they would be 1.4v somewhere... here it is in the AMDzone review:
If it takes +0.25v to get 2500 MHz and it only takes +0.1 - 0.15 on the Newcastles, that would point towards the newcastles still being a bit better wouldn't it?
TRENDING THREADS
-
Discussion Zen 5 Speculation (EPYC Turin and Strix Point/Granite Ridge - Ryzen 9000)- Started by DisEnchantment
- Replies: 25K
-
Discussion Intel Meteor, Arrow, Lunar & Panther Lakes + WCL Discussion Threads
- Started by Tigerick
- Replies: 25K
-
Discussion Intel current and future Lakes & Rapids thread- Started by TheF34RChannel
- Replies: 24K
-
-
