X-bit labs

Started by: Niels | Date 05/22/06

Comments: 24 | Last Comment:  08/25/06

[1-14]

1. WTF no overclocking??
nice review btw, but i really miss the OC part

2. page 10 lists the 5000 incorrectly. states that it has 2 x 1mb

3.
Well laid out and not too technica.l

4. Those CPUs are already running at their maximum clockspeed, nothing to overclock =) (2.8 GHz is the max)

5. "" Those CPUs are already running at their maximum clockspeed, nothing to overclock =) (2.8 GHz is the max) ""

lets see . some reviewers have already hit over 3.2ghz , 2.8 is far from the max

nice review , well done

And how did they do that?
They either used
a, aftermarket cooling
or
b, changes to Vcore
Or they had higher quality CPUs (center of the wafer)

you are one dumb fuck, really
check XS
oppainter had an FX-62 at -40° stable at 3667MHZ on 1.6Vcore
thats insane
and best of all, it was one that was baked on the outside of the wafer
your turn for some other dumb reply

6. I think there is no reason to change anything inside CPU. I expect some changes then they start to produce in 65nm, but it would be nice if they would improve at least power consumption. It is risky, because Intel is trying to scare buyers with his still phantom conroe.

7. AAAAAAAAAAAAAAAAAAaaaa I hope Intel finally dies!

Intel Conroe @ 3.9GHz: SuperPI 1M - 12.984s
http://www.xtremesystems.org/forums/showthread.php?t=99980

AMD FX-57 @ 4.2GHz: SuperPI 1M - 21.992s
http://www.xtremesystems.org/forums/showthread.php?t=100101

In other words, AMD is 2-times slower...

Please stop posting such ignorant comments. It only shows your level of îgnorance ;)



You CAN'T base off CPU performance off 1 benchmark. Sure, it's almost twice as fast, but only if all you do is run SPi all day and sit there all day long, watching the times pop on on the screen.

The K8 is almost twice as fast on some benchmarks compared to the P4 processors. Does that mean it's 100% overall better, NO!



The CD2 advantage is more like 20% per clock on most benchmarks. On some it's lower than that. The CD2 even loses some benchmarks to the K8 in certain applications and under certain circumstances.

AFAIK, it's the shared humongous cache that gives the CD2 such a huge advantage. On some benchmarks, once the calculation surpasses the 4MB mark, the K8 even takes the lead.

What's so special about the shared cache, is that both CPU's can use it. SO I think in a one threaded application ,one core can use the whole 4MB for itself, while each core in the K8 is restricted to 1MB cache, even if only 1 core is being used.

We've still got to see a thorough review on the CD2 that show's what makes it tick so fast. Time will tell.

He-man

AMD fanboys are very funny.

I own X2 4400+ myself, bought it at 6/2005, but right now AMD has the worst price/performance ratio there is. For a price of quite obsolete Venice 3700+ you can buy dualcore Intel D930 65nm and overclock it to 4.5GHz.

AMD needs some very serious spanking.. and Conroe is exactly that.

PS: SuperPI is about FPU performance, which tells A LOT about reallife performance in games, which you could know, fanboy ;)

you are really stupid
if you mention 3700+ is a venice
that proofs it

Correction: Athlon 64 3700+ Socket 939 -> San Deigo, Not Venice.
Clarification: "Venice" is in a different class than the Dual Core D930. Its a higher clocked single core CPU.
Questionable comment: You debated the "value" (Price/performance ratio) of the Athlon CPUs due to the D930's ability to overclock. A 4.5GHz D930 isn't exactly something reasonablely attainable for most people, not to mention its impractical power useage. You'd also need to pair it up with a good strong PSU and Motherboard. You also never mention the overclockablity of AMD's CPUs, along with the value-segment Motherboards that will allow you to overclock the CPUs. Athlon X2's can generally hit 2.6GHz with ease, with reasonable power consumption and power output (and this can be done on a cheap $60-70 board, like the ASRock 939DUAL-SATA). A 3700+ San Diego can hit anywhere from 2.7GHz - 3GHz, which can really give a mildly OC'd D930 a run for its money in a single threaded app. Another thing about SuperPI, its STILL a synthetic benchmark, so it really doesn't tell me much about "real life" performance. Its still a whole bunch of number crunching, not a real life simulation of common user applications. Your telling me it tells me a lot about real life performance in games. Well, so do gaming benchmarks, except those are a crud load more accurate and actually tell me *exactly* what kind of performance to expect. An sPi benchmark won't tell me what performance to expect from a CPU.

8. You might want to check the amount of power being drawn by the motherboard. I remember reading a quick review somewhere that claimed the new NV MB was drawing as much as 30watts more than an ATI based MB.

Nice article. Would love to see some OCing though.

He-man

"We measured the current that goes through the CPU power circuitry."

9. well, I think AMD knows exacty what they are doing by releasing these chips out right now. They are very futuristic in their thoughts. Since they have started the high production of these chips before now they only dim it fit to make it all good by introducing to the world the last candidate of 90nm process just to round up the playlet on this technology manufacturing process. No one knows what will before other chips that are yet to be produced from 65nm which Intel is celebrating in advance. That was the reason there was no modification at all apart from memory banwidth and 200MHz increase in clock speed. Well the whole drama is a lot more intereting now. AMD vs Intel on 65nm that is what is coming very shortly. No one knows the performance of AMD chips on 65nm yet at all. Let no Intel fan dance yet.

10. Quick question: Can one person install regular DDR memory with the AM2 cpus, or does it HAVE to be DDR2?

Socket AM2 is DDR2 only, you HAVE TO install DDR2 modules on Socket AM2 motherboards.

11. It would be helpful if your reviewer could add a one liner under the benchmark test name :
" Higher number is better or Lower number is better ".
That way even non-techies like myself can comprehend the benchmark numbers on the graph.
CNET does this on their reviews. I find it most helpful. Thanks.

12. Agree with #11

Why DDR-SDRam don't boost speed more than (200*2)?
Where the information of memory architecture, how it work; DDR2, DDR3?

13. VERY WELL

14. FX 2 5000+

[1-14]