Some thoughts on the New Procs

Dev

Distinguished
Sep 18, 2001
553
0
18,980
Well there's a lot reactions on the different reviews out there. However, I was surpised that none of the reviews I read mentioned the alpha status of the OS. Given that the benchmarks were rather close for the P4ee and the A64, I would think that this has a major impact on the performance of the A64.
While not really a good indicator due to the architectural differences, it's still worth noticing that a 333MHz Ultra SPARC beats a 1.5 gig pentium/xp in Solaris performance due to the huge hit taken by 64 to 32 bit conversion and the optimization for RISC type processor. I really don't give these early reviews much attention as we will have to wait for some optimized and stable drivers and OS for the A64, then the game is on. I don't think we'll see the "real results" for some time yet as manufacturers are still tweaking their hardware and software. just my $0.02

Dev

---
My Sig:
))
(( ___________________
|||_____________|_____|
 

Bahumut

Distinguished
Sep 18, 2002
193
0
18,680
We will have to wait a while before we see the true performance of AMD's new processors, but Intel's should be running pretty close to max.

Intel's P4EE's have little/no core design changes in comparison to their previous P4. They should've had little trouble with implementation(memory interface). Once these issues are fixed, performance will rise, but marginally.

I was interested to find out the extra L3 cache actually helped out in gaming. I didn't think it would. I'll do more homework next time.

AMD has had an extra year to prepare their processor, they shouldn't be having any problems, but none of the current solutions seem to be running as intended. I would've expected these problems a year ago, but not now. *Imagine if they would've released these processors on schedule.

While the P4EE's are being introduced nearly maxed out, the A64's have a bit of room for improvement. Memory timings have been a huge performance enhancer since the athlons came out. The AGP flaw in nVidia's solution is a huge problem that, once fixed, will improve performance nicely. Or perhaps VIA will finally come up with a dual DDR solution. AMD lengthened the A64s pipeline so they have a little more room to improve speed, though with only 12 stages I don't think they'll make it to 2.8 GHz. Maybe with the .09 process it'll be possible.

No matter how you look at it, AMD is still behind Intel, though much closer now. The problem is, they can no longer taunt Intel with price/performance ratios. With Prescott coming, AMD needs another innovation, or a way to cut production costs.

Pain is the realization of your own weakness.
 

Dev

Distinguished
Sep 18, 2001
553
0
18,980
I was surprised by the performance gain of the L3 cache as well. I think it's faily obvious that the desktop belongs to Intel at the moment in almost every category, but for those of us who can see tangible benefits from 64-bit, the opteron is a fairly good deal. If, and a big if at that, 64 bit applications become "mainstream" on the client side, or if you already have that requirement, I think the A64 makes a good choice as far as price/performance goes.

OT, but I have been bored with the CPU tech for a while, but now with the Pentium-M, Opteron, A64 and prescott it seems that we will have a lot to look forward to.

Dev

---
My Sig:
))
(( ___________________
|||_____________|_____|
 

Kelledin

Distinguished
Mar 1, 2001
2,183
0
19,780
Well there's a lot reactions on the different reviews out there. However, I was surpised that none of the reviews I read mentioned the alpha status of the OS.
Well, a lot of reviews didn't even bother to try a 64-bit OS, alpha-status or not. Ace's and Anand did, and saw some pretty interesting performance boosts.

While not really a good indicator due to the architectural differences, it's still worth noticing that a 333MHz Ultra SPARC beats a 1.5 gig pentium/xp in Solaris performance due to the huge hit taken by 64 to 32 bit conversion and the optimization for RISC type processor.
Along the same vein (and to balance things out), it's also worth noting that Solaris x86 is a fricking joke. Not even die-hard Solaris guys take it seriously. It's a piddling little afterthought of an OS Sun came up with to gain mindshare (and then left crippled in favor of Solaris SPARC).

It goes kind of like this:
<pre><ValuedCustomer> Man, I'd love to try out this Solaris thing. I just don't see the advantage of buying SPARC...
<ScottMcNealy> Here, I'll let you try Solaris x86.
* A few weeks pass...
<ValuedCustomer> Man, this Solaris thing might be cool, but it's slow and doesn't work with some of my x86 hardware!
<ScottMcNealy> What do you expect? You're running a crappy x86 system. Here, try some of our much superior SPARC stuff!
</pre><p>That's Scott's attitude. In his little mind, x86 is worthless, and nobody in their right mind should be running anything but SPARC. Solaris x86 is strictly a vehicle for Sun to try to funnel customers from the x86 world. Scott apparently (and stupidly) believes that leaving that vehicle riding on rims isn't really a significant hindrance to its purpose.

Sun only provides x86 servers because the market demanded it loudly and flatly refused to pay a fortune on SPARC boxen.
</rant>

Sorry...went off on a tangent there.

<i>I can love my fellow man...but I'm damned if I'll love yours.</i>
 

Dev

Distinguished
Sep 18, 2001
553
0
18,980
Along the same vein (and to balance things out), it's also worth noting that Solaris x86 is a fricking joke. Not even die-hard Solaris guys take it seriously. It's a piddling little afterthought of an OS Sun came up with to gain mindshare (and then left crippled in favor of Solaris SPARC).

I absolutely agree. Jeez, they didn't even bother to code in support for the power management. However, I really like Solaris on SPARC though. Rock stable, hot swappable everything and serious performance.

Dev

---
My Sig:
))
(( ___________________
|||_____________|_____|
 

coolsquirtle

Distinguished
Jan 4, 2003
2,717
0
20,780
I really can't believe how Intel beat AMD with cache~~~~ but Hammer still kills~~~ now we need a value $150 version of the hammer so i can setup a mini server/comp with it :D

RIP Block Heater....HELLO P4~~~~~
120% nVidia Fanboy
PROUD OWNER OF THE GEFORCE FX 5950ULTRA <-- I wish this was me
waiting for aBox~~~~~~~~~~~~~~~~
 

coolsquirtle

Distinguished
Jan 4, 2003
2,717
0
20,780
I really can't believe how Intel beat AMD with cache~~~~ but Hammer still kills~~~ now we need a value $150 version of the hammer so i can setup a mini server/comp with it :D

RIP Block Heater....HELLO P4~~~~~
120% nVidia Fanboy
PROUD OWNER OF THE GEFORCE FX 5950ULTRA <-- I wish this was me
waiting for aBox~~~~~~~~~~~~~~~~
 

Kelledin

Distinguished
Mar 1, 2001
2,183
0
19,780
I really can't believe how Intel beat AMD with cache
Well, P4EE 3.2 didn't quite beat it. Maybe performed on par with FX-51, but no real victory. Plus, P4EE is the epitome of a paper launch, much like the first AthlonXP 2800+ was.

Intel's essentially saying, "hey, who cares about what AMD's done--look what we can do! ...in a month or so, that is." Meh on that. If Intel can't meet AMD's game <i>today,</i> Intel's punked. Two months from now, if Intel can actually deliver what it's showing off, things may be different.

<i>I can love my fellow man...but I'm damned if I'll love yours.</i>