My Athlon XP has SSE2 !

[scrap]

Hello,

I have a PC with strange specifications.
It as been sold as Athlon XP. But it has got SSE2 instructions. AFAIK Athlon XP should not have these instruction set.
So, what is my CPU ? A Sempron, maybe ?

[root@localhost root]# cat /proc/cpuinfo
processor : 0
vendor_id : AuthenticAMD
cpu family : 15
model : 4
model name : AMD Athlon(tm) XP Processor 3000+
stepping : 8
cpu MHz : 1596.070
cache size : 256 KB
fdiv_bug : no
hlt_bug : no
f00f_bug : no
coma_bug : no
fpu : yes
fpu_exception : yes
cpuid level : 1
wp : yes
flags : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush mmx fxsr sse sse2 syscall nx mmxext 3dnowext 3dnow
bogomips : 3162.11



SoB Client detects my CPU as an Athlon XP.

So, the question is : Should SoB client detect this strange CPU ? If SSE2 core is faster than non-SSE2, then I think it should

[CaptainMooseInc]

I have heard stories of higher end chips being sold as lower end MHz/GHz chips before. Supposedly the AMD Sempron 3100+ is the same processor as the AMD64 3000+, but with some blockage to make it a 32 bit chip (kinda like that whole L5 bride that you can connect and "unlock" an XP to be an MP processor). I don't know all the exact details, so don't take what I'm saying to heart. Quite possibly you have one of those chips, or have found a bug in SoB.

Either way, if it works, don't fix it.

[Mystwalker]

Higher-end Semprons are based on the Athlon64, AFAIK. This can be (more or less between the lines) read here.
I once had a better reference, but can't find it right now.

Basically, the A64-Sempron lacks the 64-bit extensions. But SSE2 could be working...

[vjs]

This is very interesting, Would you check the client speed on your processor using both the sse2 and non-sse2 client.

Might be worth posting in some of the forums.

Details on the processor, where you puchased it manufacture date etc. If you could get the codes of the black strip next to the die that would be great WOW!!!

[Joe O]

I don't know about the Sempron 3000+, but I can confirm that the Sempron 3100+ has SSE2.

[kugano]

The SSE2 client should work. It doesn't really care about the name ("Athlon XP") of the processor... it just checks the feature flags from the CPUID instruction (that's where the "flags" line comes from in /proc/cpuinfo too). It won't disable SSE2 just because it thinks Athlon XP's shouldn't have it. If your processor is really reporting that it has SSE2 then it should work. Mais oui, c'est bizarre! Nous voudrions savoir les résultats... bonne chance!

[vjs]

Get sissoftsandra and check if it's a socket A or not.
I'm certain you have the 754-pin in which case it does have sse2.

The socket A Sempron versions looks like a bad investment, you would have been better off with almost any other socket a chip for approx the same money. But this is really only from a overclocking stand point.

[scrap]

Hello,

Thank you all for your replies.
Here are some screenshot with lastest version of sotfware.





With a week of brain storm, I concluded it is a downgraded Athlon64.
It has all of an Athlon 64 :
- CPUID check
- HyperTransport
- Powernow specific to k8 (Linux found: powernow-k8: Found 1 AMD Athlon 64 / Opteron processors)
- Has SSE2
- But has no 64 bit registers !!!! That's the only difference !

If anyone has another conclusion, I'm ready to hear

[CaptainMooseInc]

Typical AMD for ya. Let's make a bunch of chips and remove some stuff and sell them as a lower-branded chip. BOOO!!!!!

Something needs to be done about this!

-Jeff

[vjs]

What are you kidding?

Let's make a bunch of high dollar 512Kb cache 2500 mhz chips that are SMP capable.
Cut a few bridges, lower the voltage, and sell them as non-smp, 1833 mhz chips, now sell them for 80 bucks.

Buy them for 80 bucks and you get what was advertised. Buy an additional 5 dollars of conductive paint 10 minutes of research and you have chips that walk circles around intel chips at double the price and could be used in SMP machines.

Look at the the, P4EE, Xeon DP, Xeon MP, all basically the same chip but there is no way of modifing them. Even a different socket for DP and MP processors. Sure don't support it but why stop people from modifying.

There are rumors that all P4's have HT... If it were AMD there would probably be a brige you could connect to enable it. I'd actually complain about the opteron not being the same pin outs, and what's the deal with the 1XX, 2XX, 4XX, 8XX, we all know they were intended to be 8XX just validate them at whatever and we will take our chances.

[SkullHead]

That is interesting. The closest thing I can see is the AMD Sempron 3100+ 1.8GHz, 256KB L2 Cache Socket 754 Processor - OEM from the specs at newegg. Sounds like if you bought a complete system someone made a mistake in calling it an XP. I'd like to see a pic of the chip to see what the die looks like. If it really is a sempron, which I think it is due to the L2/256 cache and the rest of the specs, then you can take it up to 1800Mhz according to the specs below. A Paris core eh? Although I can't figure out where the XP comes from in the sisoft sandra

• Model: AMD Sempron
  Core: Paris
  Operating Frequency: 1.8GHz
  FSB: Integrated into Chip
  Cache: L1/64K+64K; L2/ 256KB
  Voltage: 1.5V
  Process: 0.13Micron
  Socket: Socket 754
  Multimedia Instruction: MMX, SSE, SSE2, 3DNOW!, 3DNOW!+
  Packaging: OEM(Processor Only)

You might want to go into the bios and see if you can bump it up another 200Mhz.

[MerePeer]

For reference, here's my {normal} 3000+

~:$ cat /proc/cpuinfo
processor : 0
vendor_id : AuthenticAMD
cpu family : 6
model : 10
model name : AMD Athlon(tm) XP 3000+
stepping : 0
cpu MHz : 2101.380
cache size : 512 KB
fdiv_bug : no
hlt_bug : no
f00f_bug : no
coma_bug : no
fpu : yes
fpu_exception : yes
cpuid level : 1
wp : yes
flags : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 mmx fxsr sse syscall mmxext 3dnowext 3dnow
bogomips : 4153.34

[maddog1]

@MerePeer:The numbers you mention for your 3000+ (Barton core-400 FSB) are correct. I have to note though that there is another version of the processor: Mine has the exact same characteristics, except it works at a slightly higher 2162,72 MHz, which is correct IMO, since it is roughly 166,6x13 (FSBxMultiplier)
Actually there were 2 processors released as XP 3000+ :
-400 FSB/10,5 multiplier->200x10,5=2100 MHz (like yours)
-(slightly older)333 FSB/13 multiplier->166,6x13=2167 MHz (like mine)

Of course, for what this thread is worth, none of those come even close to the processor Scrap describes.
A pretty good table with mostly every AMD processor specs is here:
http://www.thedigerati.us/info/amdcpuchart.html

While I agree that the chip in question is 99% a Sempron and it *could* be a downclocked 3100+ as described by SkullHead another good candidate seems to be the Sempron Mobile 2800+ described here
It seems like the only perfect match (socket type, frequency, multiplier, L2 cache)
That is, assuming the programs used return the most basic data correctly...

Of course, the only way to know for sure, is if Scrap finds out the cpu id tag, as printed on the processor itself. For example, my 3000+ is actually the "AXDA3000DKV4D" using AMD's codes.

Anyone has any better ideas? AMD seems to have a PDG mess of processors available!

[meckano]

A smaller list I found long time ago.
only goes up to barton.

http://www.cpuscorecard.com/cpuprices/aaxp.htm

But shows cpu's not on other list, like TBred 2800+