90nm A64: Value, Speed and Overclocking(Air) REVIEW / Q&A with AMD

There is one thing that really pisses me off about that review. After they saw that both the 3500+ 90nm and the 3000+ 90nm both reached the same OC speed, they ran the 3500+ 90nm at that speed (290x9) yet did not run the 3000+ at that speed, but instead ran it at stock speeds for the benchmarks. Whats up with that? Or am I just seeing it wrong?
 
EmptyWallet said:
There is one thing that really pisses me off about that review. After they saw that both the 3500+ 90nm and the 3000+ 90nm both reached the same OC speed, they ran the 3500+ 90nm at that speed (290x9) yet did not run the 3000+ at that speed, but instead ran it at stock speeds for the benchmarks. Whats up with that? Or am I just seeing it wrong?
Click to expand...

They say .09 3500+ /3000+@290x9 on their graphs and at the very bottom they have comments like "comparing benchmarks of the 3000+ at stock speed to the same benches at 2.6GHz (290x9)." so id guess they used both and they ran the same so they combined them. who knows and the 3500+ and 3000+ were bought from different vendors dunno if that matters or not? =/
 
EmptyWallet said:
What MB were they using?
Click to expand...

MSI K8N Neo2 for all of the 939's with Thermaltake Silent Boost K8 Heatsink/Fan for the Overclocking part
I find it interesting that all their 90nm parts are performing at stock better than the 130nm ones in their testing
 
Yeah saw it - reading now...

Awesome - looks like it confirms the other review on Madshrimps - definitely getting a 3000+.

One question though. I thought that the new 90nm chips came in 3000+ 3200+ and 3400+? That review kept saying 3500+ instead of 3400+? Oh well doesnt matter, 3000+ is the sweet spot here.

Just need.... N...Force......4..................
 
I wish they reviewed a socket 754 chip in there too at 2.5 or 2.6ghz :(

I want to see how my Athlon S754 @ 2.6 performs against the new 90nm.
 
As AMD plans to have 50% of their chips be 90nm by the end of the year, i would guess they would be.
 
Looks like I am building my last NF7s and Mobile 2600. Finally I have a good overclocker HOPEFULLY with the 3000+
 
Woot!! im definetly gonna look into these processors.... btw why wouldnt you guys get a 3200 939 .9 micron Processor?? its a 10 Multiplier and 200 HTT Defaults .... i think it would be even easiar to get 250 HTT x 10 and get 2.5 ghz :D :D
 
Dudeyourlame said:
Woot!! im definetly gonna look into these processors.... btw why wouldnt you guys get a 3200 939 .9 micron Processor?? its a 10 Multiplier and 200 HTT Defaults .... i think it would be even easiar to get 250 HTT x 10 and get 2.5 ghz :D :D
Click to expand...

cause 280x9 sounds oh so much nicer :D

ill probably pick up the 3200+ when i buy, then i can choose either the 10 or the 9 multi.
 
Isnt the 3200+ like $240 or something? Thats breaking the $200 marker. Cant have that...
 
The parts they got up to 2.6, but there is no way AMD would sell those as 2.6 parts. You'd figure at least a 5-10% margin for error/outside circumstances (hot room, crappy case, stock HSF and voltage, etc) -- you'd need to see somewhat common OCs of 2.75 before they put a 90nm 2.6 out there. Its possible before the end of the year if they can squeeze it out, but I'm guessing a new stepping after Jan 1 for any 90nm parts 2.6 and faster. For now, 130nm can get to 2.6 readily enough to be shippable.
 
Got my 3000+ 939 today. It is late and I won't do some serious stress testing until later, but I match thier results, and I have not tried any higher. The picture below is probably small and hard to read, but here is a summary:

A64 3000+ s939 @ 2610 9x290
MSI K8N Neo2
Kingston HyperX PC3200 @186mhz at nominal timings
56 degrees C after a few runs of Sandra Burn in (40 idle)
GF6800GT @default speeds

51a4.jpg


I haven't quite understood why a lot of folks were hyping the 3200+ over the 3000+, it appears as though it is the usual; the lowest end will o/c as much as the mid/high end with the s939s.

Stability hasn't been verified completely, as it is late and I just wanted to get an idea of the o/c potential of the 3000+. That is also why I have the memory so low. I'll tweak it out later. I got through about 6 runs of sandra burn-in and ran 3dmark2k1 a couple times. I'll do prime later.

Looks good so far, though.
 
11days, you have stock HSF? The advantage of the 3200 is that people like me are hoping that because of the x10 multiplier we could get up to 2.9-3.0 with nice cooling. Also, note in the anandtech review that the 3500 got to 2.61 at stock voltage, while the 3000 had to go to just over 1.5v, so maybe the 3200 can actually oc a little better with increased voltage, but anandtech didnt feel like pushing it.

This like many posts is just speculation, however, as it is possible that there is a serious 2.6-2.65 roof on the 90nm, in which case, yes the 3000 is definately the best deal.

Also, Everyone should note that the 3000+ 90nm at 2.6 is VERY SIGNIFICANTLY faster than the FX-53 stock. Now after that why amd is making the 4000+ on 130nm is beyond me...

Also, 11 days your link isnt working for me.
 
ToyYoda03 said:
I wish they reviewed a socket 754 chip in there too at 2.5 or 2.6ghz :(

I want to see how my Athlon S754 @ 2.6 performs against the new 90nm.
Click to expand...

Madshrimp's review of the same part did this I believe, look for it, was posted on the [H] frontpage a couple days ago.

9mmx19 said:
One question though. I thought that the new 90nm chips came in 3000+ 3200+ and 3400+? That review kept saying 3500+ instead of 3400+? Oh well doesnt matter, 3000+ is the sweet spot here.

Just need.... N...Force......4..................
Click to expand...

The 3500+ was/is the 130mm Newcastle s939 part at 2.2ghz. There's some s939 3400+s floating around (previously 3400+ was used for the 2.4 s754 part, I think) that are apparently just 3500+s on the same 130mm process but with a 800HTT instead of a 1,000 HTT. There's two threads about this on the boards...

Not sure why'd they release such a processor but all the evidence points to this. It sells for slightly less than the 3500+. Lastly... What's the obssession with nForce4, I don't see what it offers other than the option of SLI/PCIe which you're gonna pay thru the eye for since PCIe boards are scarce and you'll see no performance gain ('least not 'till Nvidia officially releases SLI).

Oh and yea, the s754 3000+ is going for $165 at Newegg, the s939 is going for $190 (it'll probably come down within a few weeks) and the s939 3200+ is going for $235. I'm frugal and I'd like a challenge so I'd go for the 90mm 3000+ just for the heck of it, I seriously doubt they'll all be hitting 2.6ghz tho (800mhz OC is far from normal for A64s these days, 90mm notwhitstanding).
 
Started my post before 11DayZ had made his... Those are some really encouraging results tho, it's quite possible the 90mm process has let a lot of possibly higher-rated parts filter thru to lower bins just to apease demand since the focus of the process is so narrow right now.

The reason AMD is still going ahead with the higher rated parts on the old process is probably because it's been tested and the results are more predictable by now. Maybe they just haven't prepared their 90mm process for higher speed rating... Or they want to get the lower-rated chips out on this process first since that's where they'll make the bulk of the sales and where they can save the most by using the process.

Whatever the reason is you can bet it's due to monetary gain and/or refinement of their process, not just someone's fancy. :cool:
 
Alright!!! I just purchased the last of my components last night. Here's what I'll be running.

s939 90nm 3200+ with XP90/Panaflo M1A
MSI Neo2
OCZ PC3700EB 1GB (2x512)
Fortron 530w PSU
ect. ect.
I wonder how high I'll be able to get with that PC3700EB?
This will be the FIRST system I've OC'd since my old T-BIRD 1.2 So if you guys don't mind then I'll be needing a little help, as I don't understand diddly about the HTT and such. I just remember when it was only raise the FSB and Multi and off you go. I suppose there is an FAQ or some such around here I could dig into, I'll look.
 
emptywallet,

some people report problems running the ocz 3700eb with the neo2. be careful.
 
i just ordered the:

3000+ .09
neo2
ocz 3200 v2 platinum 2x512
and the same fan they used in the review, the thermaltake silent boost

crossing my fingers for a big overclock!
 
stelleg151 said:
11days, you have stock HSF? ... Also, 11 days your link isnt working for me.
Click to expand...

I'm using a coolermaster hyper6, one of the tower heatpipe jobs.

 
SPARAT1K said:
i just ordered the:

3000+ .09
neo2
ocz 3200 v2 platinum 2x512
and the same fan they used in the review, the thermaltake silent boost

crossing my fingers for a big overclock!
Click to expand...

I'm getting the same hardware monday. Will keep you posted.
 
You must log in or register to reply here.
Back
Top