• 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.

New Intel 6xx series available

To answer my own question, I found a review of a 660 engineering sample here. The review is in Chinese, but the performance stats seem clear enough. Looks like you can expect a 1-10% performance improvement over the 560 depending on your application. And I assume that doesn't factor in the improvements that can come with 64-bit code:

http://www.hkepc.com/hwdb/p4660-5.htm
 
Check out the Xeon review on the front page of AT. That should give you a good idea of the performance gains.

Im rather impressed 2mb L2 cache did this much.

-Kevin
 
I suspect that those three applications can all benefit greatly when their code is optimized for 64-bit. That may take some time though. The 2MB L2 will probably help too, but to a lesser degree than 64-bit optimizations.
 
Has anyone purchased one of these yet? Has it shipped? Just wondering before I pull the trigger, I want to get a SFF 64 bit system up and running with W2K3 server and the P series shuttle will let me have 3 hard drives in it. Unfortunately, no P series for the Athlon 64 yet and I'm tired of waiting.
 
Ok bought a 630 today, default 3.0GHz, overclocking stably at 3.75GHz using Asus P5GD2 and running DDRII at 667. Temps are constantly at 52 degrees centigrade while running CPU burn-in.
Can't go over this freq as my WD Raptor SATA can't be recognized at post, anyone have any ideas how to get around this using the Asus Bios? The multiplier is totally locked at 15x and the Asus bios doesn't seem to allow any changes to this at the moment. Is there a way to lock the SATA freq at all?
 
Originally posted by: christopherzombie
WOW, I'll bet it's close to 3.4EE performance... and for only $435!
Click to expand...

Dream on! There's a reason Intel is bieng real quiet about this chip. Expect 1-4% over current presshot.

I bet 3400 ($200) will beat the 3.6 ($702) in majority of tests.
 
I'd think you'd get a pretty good performance boost by having 2MB of cache especially in audio and video apps.

I wonder how many audio tracks one can run in cubase with one of these.
 
Originally posted by: modalone
I'd think you'd get a pretty good performance boost by having 2MB of cache especially in audio and video apps.

I wonder how many audio tracks one can run in cubase with one of these.
Click to expand...


Actually I remember this coming up awhile back when Intels only answer to the A64 was to jack of the cache in the EE model....

If I remember corretly the nature of audio and video as well as its type of data made ithe added cache a non factor.....games seemed to like it more then those type of apps....Again if I remember correctly from back then....
 
I take it that the 2mb on die cach is proper L2 cach not like the old extreme edition chips where they were using in affect SD ram for the L3 cach, which actually runs at a much lower frequency, therfore not applying great performance benefits as you might think ????
 
There are like 6 reviews floating around now, and they pretty much agree that the 6 series is no faster than the 5 series. The cache simply doesn't help as much as the 200MHz you're giving up. For example, a 640 is priced like a 550, but the 550 will outperform it pretty much around the board. The cache is like a half of a speed bump, comparable to 100MHz or so. This is good for intel, because they basically got one more speed bump out of a core that topped out at 3.8GHz. It's bad for customers, though, because you get less speed for your money.

Of course, the features are better. The power management is much better, and even under load the new CPUs eat up much less power than their 5xx cousins, which is surprising considering the 50 million or so transistors added for the new cache. And the EM64T and NX bit support are definitely nice. If I were getting a P4, I'd probably get a 6xx series, because of these features. Plus, if overclocking is good, then you can make up for the slower speed that way. Then again, I'd probably rather have an A64, considering they have all the good features (including SSE3 very soon) and performance, rather than 1 or the other.

Oh, and it looks like the 3.73GHz P4EE is no faster than the 3.46, and is actually slower sometimes due to the higher latency of its L2 cache and deep Prescott pipeline. Intel really just needs to give up the whole "EE" thing. At least the FX is the best A64 around the board. The P4EE loses to its predecessors and many mainstream P4s in a lot of benchmarks, making the $1000 chip pretty worthless.
 
I'm worried under CPU-z I do not see the multiplier of my 630 3.0G cpu as being variable between 14-15 and it doesn't drop into power saving mode either. My motherboard Asus P5DG2 Premium already has latest version Bios installed. Anyone know what's going wrong, I see under CPU-Z that the name of the CPU should be appended by ES but my omits that all other model numbers, rev etc are the same though. Are there any patches I need to install?
 
Originally posted by: Duvie
Originally posted by: modalone
I'd think you'd get a pretty good performance boost by having 2MB of cache especially in audio and video apps.
I wonder how many audio tracks one can run in cubase with one of these.
Click to expand...
Actually I remember this coming up awhile back when Intels only answer to the A64 was to jack of the cache in the EE model....
If I remember corretly the nature of audio and video as well as its type of data made ithe added cache a non factor.....games seemed to like it more then those type of apps....Again if I remember correctly from back then....
Click to expand...
That's very true. Added cache helps in a lot of scientific synthetic benchmarks, which helps Intel present the "speed" of their CPUs to the world at those design conferences, but in terms of "streaming multimedia data" - adding more cache helps practically zero there, above a certain minimal amount that allows it to keep the pipeline full of data (with the help of the hardware prefetcher unit and software prefetch "hint" opcodes).

The sad irony is that things like media-encoding are the area in which the P4 tends to excel against possible competition, and therefore, jacking up the amount of cache generally gives no benefit for those apps, and yet adds to the heat and the chips' cost significantly. Can you say, "a step backwards for Intel"?

For things like servers that manipulate in-memory databases, or scientific apps which operate between large blocks of X (op) Y data, it could be a win. But otherwise.. it seems like a desperate last gasp for performance out of the "dying" netburst architecture. (It would also be a major win for multi-processor SMP servers, because of Intel's likewise semi-brain-dead design that has all CPUs sharing a single bus to main memory. Adding L2 cache and intellegently doing load/store prefetching/post-writing of data will significantly lower contention for access to main memory. However, it may make cache-snooping between CPUs operating on shared data (locks, etc) more expensive, since there are larger cache arrays to look through on each CPU in the system.)
 
You must log in or register to reply here.

TRENDING THREADS

Back
Top