Amd 64 FX-57

  • Thread starter Thread starter potnisanish
  • Start date Start date
P

potnisanish

I know theres a 4800 amd chip but I'm wondering what the fx-57 is
equivalent too. Like, the 4800 is equivalent to an intel 4.8 GHz
processor. Does anyone know?
 
I know theres a 4800 amd chip but I'm wondering what the fx-57 is
equivalent too. Like, the 4800 is equivalent to an intel 4.8 GHz
processor. Does anyone know?

You should really be using benchmarks, rather than the manufacturer's
product codes.
 
I know theres a 4800 amd chip but I'm wondering what the fx-57 is
equivalent too. Like, the 4800 is equivalent to an intel 4.8 GHz
processor. Does anyone know?

The 4800+ is a dual core, the FX-57 is a single core chip. The 4800+ runs
at 2.4GHz the FX-57 at 2.8GHz. Both chips are overpriced. The FX-57 is
$1011 on Pricewatch, the 4800+ is $884. For dual cores the chip of choice
is the 4400+, it's only 9% slower then the 4800+ but it 40% cheaper
($548). The A64 4000+ is a single core chip that runs at 2.4GHz vs 2.8
which is 15% slower then the FX-57 (the cache size is the same as the
FX-57, 1M, so the only difference is clockspeed). The price however is
$368, 64% cheaper.

You can't really compare AMD's number to Intel's clock rate on a one to
one basis. On many applications the AMD number is extremely conservative,
on others it's about right. For example my benchmarking doing Verilog
simulation finds that the A64 3400+ is equivalent to a 5.2GHz P4. Verilog
is an example of an integer application, operating systems and word
processors are also integer codes. Integer codes are very sensitive to
things like pipeline depth and memory latency, both areas where the Athlon
64 design is much better then the P4 design. Things like games and
multimedia encoders contain vectorizable floating point code, which in the
x86 world are done with the SSE3 instructions. SSE3 instructions are
insensitive to pipeline depth and benefit from high clock rates and memory
bandwidth, two areas where the P4 is superior the A64. However even game
contain a lot integer code so the P4s superior SSE3 performance is
balanced out by it's lousy integer performance. The bottom line is that
the AMD rating number is crudely equivalent to Intel's clock rate as a
means of comparing single cores. If you bought a 4000+ it would run
approximately like a 4GHz P4 (+/-15%) on things like games. It would also
run a lot cooler.

Dual core vs single core is a separate discussion. If your running
multiple applications simultaneously or if your running multithreaded
applications then the dual cores are a huge win. If you are primarily
running one single threaded application at a time then dual cores give you
a very small improvement, it's not 0 because the operating system always
has daemons running in the background that can use some of the available
computing power, but it's small. Dual core systems generally feel more
responsive because even if you have one cpu bound application running you
have plenty of additional capacity remaining so there is a reason to buy a
dual core even if what you are doing right now is primarily single
threaded. Between AMD and P4 dual core chips there is no contest. The A64
was designed for the purpose for the beginning, the P4 wasn't and it
shows. There have been some comparisons on the hardware sites. The figure
that sticks in my mind is that the Athlon 64 X2 delivers nearly 2X the
throughput of the equivalent single core A64. The dual core P4 delivers
only 1.6X which is awful.

Bottom line, the best choices for a high performance system are the A64
4000+ for single core systems and the A64 X2 4400+ for dual core systems.
 
Between AMD and P4 dual core chips there is no contest. The A64
was designed for the purpose for the beginning, the P4 wasn't and it
shows. There have been some comparisons on the hardware sites. The figure
that sticks in my mind is that the Athlon 64 X2 delivers nearly 2X the
throughput of the equivalent single core A64. The dual core P4 delivers
only 1.6X which is awful.

Bottom line, the best choices for a high performance system are the A64
4000+ for single core systems and the A64 X2 4400+ for dual core systems.

I have a dual core P4 3.0G and it performs just as I expected/hoped. For
the jobs that I run (hours of regression tests), the dual cores provide 2x
the processing of a single core (about 2.5X the processing power of a single
2.4G P4). So, I certainly don't see the 1.6x problem. That does not mean
it doesn't exist in other situations, though. It is possible that my
applications are more CPU (rather than memory/disk) intensive and hence work
nicely on PD. I wanted to get a dual core AMD for performance and power
reasons, but they were much more expensive. Maybe next time.

As you mentioned, the dual core chips provide excellent responsiveness under
normal conditions. I was/am amazed at how poorly Windows XP handles
multi-tasking compared to UNIX. With a single core, your machine pretty
much grinds to a halt if you run a single CPU intensive job at normal
priority. The Windows guys should take a look at the Linux code!

Peter
 
Back
Top