AMD 64 socket 939 aint doing well ...

[soulkeeper]

hi

i used to have AMD 64 3200+ but then i changed to Nvidia a8n sli deluxe so the sockets was different (939) so ihad to get another cpu so i ordered 3500+ , the seller was nice and he offered me to use an AMD 64 3000+ socket 939 till he brings me the 3500+ so i ran a serious of tests ...

when i had 3200+ the Sisoftsandra used to give me about 10000 MIPS knowing that the reference cpu which is also 3200+ was giving 9200 MIPS .
now this 3000+ cpu is giving me 7600 MIPS on sisoftsandra !!!! omg how come ? even the reference CPU when i put to 3000+ is giving 9000 !!!!


so my question in SisSandra the references are :

AMD Athlon 3000+ = 9000 MIPS
AMD Athlon 3200+ = 9142 MIPS

the different between them is 142 MIPS !! why does my new 3000+ giving me less than the 3200+ ??? its about 1500 MIPS lol ? is it the new sockets ? even when i put asus N.O.S to 8% its giving me 8142 !!! should be 9000 without overclocking !!!

i mean i know its temporary cpu to me but i hope that my 3500+ will be different in speed comparing to the 3200+ and other !!!
is there something wrong i have ?

my system :

Processor
Model : AMD Athlon(tm) 64 Processor 3000+
Speed : 1.81GHz
Model Number : 3000 (estimated)
Performance Rating : PR2712 (estimated)
Type : Standard
L2 On-board Cache : 512kB ECC Synchronous, Write-Back, 16-way set, 64 byte line size

Mainboard
Bus(es) : PCI PCIe USB FireWire/1394
MP Support : 1 CPU(s)
MP APIC : No
System BIOS : Phoenix Technologies, LTD ASUS A8N-SLI DELUXE ACPI BIOS Revision 1004
System : System manufacturer System name
Mainboard : ASUSTeK Computer INC. A8N-SLI DELUXE
Total Memory : 1GB DDR-SDRAM

Chipset 1
Model : Advanced Micro Devices (AMD) Athlon 64 / Opteron HyperTransport Technology Configuration
Front Side Bus Speed : 2x 1005MHz (2010MHz data rate)
Total Memory : 1GB DDR-SDRAM
Memory Bus Speed : 2x 200MHz (400MHz data rate)

Video System
Monitor/Panel : F900B
Adapter : ASUS Extreme AX850 XT Platinum Edition
Adapter : ASUS Extreme AX850 XT Platinum Edition Secondary

Physical Storage Devices
Removable Drive : Floppy disk drive
Hard Disk : NVIDIA STRIPE 149.04G
Hard Disk : NVIDIA STRIPE 152.66G
CD-ROM/DVD : HL-DT-ST DVDRAM GSA-4160B (CD 79X Rd, 31X Wr) (DVD 10X Rd, 4X Wr)
CD-ROM/DVD : PIONEER DVD-ROM DVD-106 (CD 40X Rd) (DVD 5X Rd)
CD-ROM/DVD : TEAC CD-W540E (CD 48X Rd, 40X Wr)
CD-ROM/DVD : CD-ROM Drive
CD-ROM/DVD : CD-ROM Drive


Power Management
AC Line Status : On-Line

Operating System(s)
Windows System : Microsoft Windows XP/2002 Professional (Win32 x86) 5.01.2600 (Service Pack 2)

plz help . thanks

 

[Spitfire_x86]

It's not underforming that much. Because your 3000+ runs at 1.8 GHz. Sandra baseline 3000+ and 3200+ probably ran at 2.0 GHz (3200+ had 1 MB L2 cache)

------------
<font color=orange><b><A HREF="http://www.mozilla.org/products/firefox" target="_new">Rediscover the web</A></b></font color=orange>

 

[Ned_Flanders]

My AMD64-3200+ gets over 11000MIPS.



<pre>But it is overclocked to 2.45Ghz</pre><p>

______________
Stranded on Death Row.

 

[fishmahn]

That is probably because the baseline 3000+ is based on the 754 3000+, which runs at 2.0ghz. Yours runs at 1.8. See how you stand up against the 2800+ baseline.

You'll have a similar issue with the 3500+ if the baseline uses socket 754 baselines. When AMD came out with the 939, they tweaked the PR ratings (3000+, 3500+, etc.) and said that the add'l performance of dual channel warrants a bump of 200 over the 754 of the same speed. (i.e., the 754 2800+ runs at 1.8ghz, the 939 that runs at 1.8ghz is the 3000+) Dual channel doesn't give that much of a speed improvement so it turns out that a 754 3000+ outperforms a 939 3000+ by a little.

Mike.

 

[Rugger]

Ned - You on stock cooling with your 939? You folding with that OC?

__________________________________________________
:tongue: <font color=red>Have you read the FAQ? Searched for other posts on this topic?</font color=red><P ID="edit"><FONT SIZE=-1><EM>Edited by Rugger on 03/03/05 11:11 AM.</EM></FONT></P>

 

[soulkeeper]

so do i understand that sockets 939 performance is kinda slower than 754 when having the same model nr ex 3000+ ? so 3500+ soc 754 faster than 3500+ soc 939 or ?

maybe im n00bing ... sorry.

The Keeper was here

 

[Guest]

Nah, you're not n00bing, you're dead on

<font color=green> Woohoo!! I am officially an <b> Enthusiast </b>!! </font color=green>
<i> <font color=red> One new Firefox fan </font color=red> </i>

 

[endyen]

For a lot of applications, a s754 3400+ is faster than a s939 3500+. That is even more true, if the 3500+ is newcastle (130 nanometer).
Just the same, s939 is dual core ready, and available with PCI-exp. Not that I would give a plug nickel for bothe.

 

[Ned_Flanders]

No, unfortunatly I'm on water. Makes my o/c look small don't it? :lol:

And yes, I'm folding away with it.

Idle - 35c, folding load 42c (1.6Vcore)

______________
Stranded on Death Row.

 

[fishmahn]

Yep... Now if AMD hadn't messed with the PR ratings, 939 would have given a net performance INCREASE - keep the 2.0ghz 939 cpu at a 3000+ (or even call it a 3100+ if you have to show it's faster) instead of calling it a 3200+ (meaning it is compared to the old 3200+ - and it doesn't make the cut then...) and they'd have gotten their faster performing chip, and a much easier sell on 939.

Depending on what you need, 939 is still a good socket, because that's where the future is; it has dual channel, PCI-e support (though I understand there is talk of some 754 PCI-e boards), supposedly dual-core will run on it as is (maybe a bios update), and the winchester A64's are supposed to overclock to higher limits, meaning they'll outperform the 754 counterparts at the extreem end of things.

Mike.

 

[Rugger]

:lol: Of course it's not insignificant - especially with those temps!

<pre>Keep telling yourself that you <b>aren't</b> inadequate</pre><p>
Really that's a good solid OC. You can be confident that you can push it further at will!

__________________________________________________
:tongue: <font color=red>Have you read the FAQ? Searched for other posts on this topic?</font color=red>

 

[pat]

Yep... Now if AMD hadn't messed with the PR ratings, 939 would have given a net performance INCREASE - keep the 2.0ghz 939 cpu at a 3000+ (or even call it a 3100+ if you have to show it's faster) instead of calling it a 3200+ (meaning it is compared to the old 3200+ - and it doesn't make the cut then...) and they'd have gotten their faster performing chip, and a much easier sell on 939.

ok... a p4 2.4a is slower that a p4 2.4c.. the difference is on the FSB 800 vs 533(or 400???, anyway), but they are still p4 2.4.. a and c is what is different. So why making noise about 3200+? put 939 and 754 to do as intel with their A or C notation. Ok, Intel doesnt make A processor anymore, but what about the E.. that is somewhat slower than the C.. OMG.. now thay have funky number meaning absolutly nothing for now.. 340, 560,.. and Celeron D..

Let the Socket A dissappear. You'll have the 754 and 939. But with Athlon and Sempron for 754. So what. Then You'll have the Opteron. Ok, if you ever need them, you are supposed to know enough about computer to understand what they mean.

So, if someone want a CPU, from AMD, yous explain them the difference between Athlon and Sempron. If he choose the Sempron, you told them about which want he want, the smallest number being the slower and going up. If he want a Sempron, then he need e cheap computer, so he will take the cheapest one in the majority of the time.

If he choose the A64, then he must understand the difference about each socket, and that the job of the adviser. Easy enough to figure. then no matther which one he choosed, then you show him the choice he has and will usually take the faster he can afford. I dont see what is difficult there?



-Always put the blame on you first, then on the hardware !!!

 

[fishmahn]

ok... a p4 2.4a is slower that a p4 2.4c.. the difference is on the FSB 800 vs 533(or 400???, anyway), but they are still p4 2.4.. a and c is what is different. So why making noise about 3200+? put 939 and 754 to do as intel with their A or C notation. Ok, Intel doesnt make A processor anymore, but what about the E.. that is somewhat slower than the C..

Intel's naming convention has always been based on clock speed (until recently, of course). A/C/E is differences in the core. AMD's naming convention has been based on a "PR" or 'performance rating' (or whatever they call it) since approx the debut of the P4, to counter the massive clock speed (but not performance) advantage Intel has.

The probable (though never official of course) reason for the 200pr increase between 754 and 939's of the same clock instead of something smaller that would have let us compare the 2 lines (754/939) as if they were one is the change in it's competitor at Intel. The 754's primary competition was the P4c (the same way the Athlon XP's primary competition was P4a - though once again, the PR rating was 'of course' not meant to compare that way), while the 939's is the P4e. So, that 'stretch' of the PR rating is valid when taken in that context because an 'X'ghz P4c isn't the same performance as a P4a or a P4e, even though they clock the same speed.

I dont see what is difficult there?

I may be misunderstanding your point here, but if I am correct, what I mean by an easier sell on 939 isn't the difficulty of explaining the differences, but the switch from 754 to 939 as the primary platform for the A64. It seems to me that for the last year or so, socket A has been the 'bargain platform'. Now, with 939, it seems that 754 should be the 'bargain' platform. However, the 3rd fastest CPU (best performing) in the whole A64 line is a 754 chip. And the top 2 are rather high-priced. That makes for a harder sell that 939 is the right platform to purchase today.

Now it's possible that AMD doesn't want a wholesale switchover from 754 to 939 immediately for its mainstream platform, in which case they're doing it the right way.

It just seemed to me that 939 doesn't come out smelling too good in light of the fact that a 754 3xxx outperforms a 939 3xxx. Let me rephrase that - it could have come out smelling better given a more consistant 'PR' naming convention. Once its understood that the pr ratings aren't meant to compare against each other outside of a 'line' (Athlon XP, Sempron, A64skt754 and A64skt939 each being a separate line), and it should be compared to its P4 competition (regardless of what AMD says - that's generally what the public would want to compare it to - XP to P4A, Sempron to CeleronD, A64/754 to P4C and A64/939 to P4E), then the pr ratings of the 939 series makes sense.

Now maybe I like looking at things with too simplistic of a view, but the confusion of the moving target PR ratings (even though there is a rationale for them) seems to me to only make things more difficult than they need to be.

I'd rather not even broach the subject of Intel's new naming convention - it seems to me it makes AMD look better... (at least they're naming their chips based on something resembling a performance # - except the FX series of course - but that can be excused as the no-holds-barred performer) - as I haven't looked closely enough at it to make sense of it...

Mike.

 

[endyen]

My take is that s939 is needed for dual core. Amd doesn't care all that much about the current chips. They just need the platform. It helps to have the bugs worked out, before the next big step. Since Amd cant supply the platform themselves, they need to get the chipset makers going.

 

[Ned_Flanders]

lol, unfortunatly this chip is at its limit. I have to coax it (with extra voltage) to turn on!

Problem is, when my room gets warm (yesterday, It was roasting), the chip hit 47 with folding! That's way higher than the 42 I have seen before.

______________
Stranded on Death Row.

 

[fishmahn]

All very true, and you're probably right. Dual cores will need the bandwidth that 939 offers. I hadn't thought about the why of 939 in the first place (it is there, the why is immaterial or something like that), just the why of 939v754 not fitting a direct comparison.

Mike.

PS: I would think it's more that AMD *won't* supply the platform themself, but in either case, they don't, so that 'why' is immaterial as well.

 

[jmwpom3]

What they needed to do was give the same #'s for the same clock speed and left the 939's extra mem support out of the equation. Then the benchmarks would've been (or should be) the same or better with the 939. (if you care about the benchmark tests anyways)So, perhaps there isn't a problem except with the benchmarking software.
But, then they couldn't have CHARGED as much for each up step.
just my opinion,

PLEASE NOTE: Some quantum physics theories suggest that when the consumer is not directly observing this product, it may cease to exist or will exist only in a vague and undetermined state. <P ID="edit"><FONT SIZE=-1><EM>Edited by jmwpom3 on 03/05/05 08:35 PM.</EM></FONT></P>