Athlon 64 questions

[Mike B]

My new athlon XP system just got fried..

I have some questions. I've heard the Athlon64 3200+ can beat a P4 3.2 EE in
most tests. Is this true, and is there any reason why one would buy the P4
over the A64 other than bias?

Also, i'm a bit confused now.. I see A64 3200+ listed at pricewatch.. some
are listed as 1600mhz, and some listed at 2.2ghz. is the 1600mhz just
describing the fsb? and do they all run at 2.2ghz?

Lastly, i see there are some that have 1mb of cache as apposed to 512 for
around the same price. Is there a catch that i'm missing?

 

[George Hizinwak]

My new athlon XP system just got fried..

I have some questions. I've heard the Athlon64 3200+ can beat a P4 3.2 EE
in
most tests. Is this true, and is there any reason why one would buy the P4
over the A64 other than bias?

Also, i'm a bit confused now.. I see A64 3200+ listed at pricewatch.. some
are listed as 1600mhz, and some listed at 2.2ghz. is the 1600mhz just
describing the fsb? and do they all run at 2.2ghz?

Lastly, i see there are some that have 1mb of cache as apposed to 512 for
around the same price. Is there a catch that i'm missing?

Mike, your best bet would be to get an AMD XP 3500 939pin cpu, with a 939pin
based motherboard. This way you can get up to 4gigs of DDRAM without losing
speed. This combo uses Dual memory channel while the 754 design uses single
memory channel and thus when you use the 3rd DIMM slot the memory slows
down.

 

[Derek Baker]

Mike, your best bet would be to get an AMD XP 3500 939pin cpu, with a
939pin based motherboard. This way you can get up to 4gigs of DDRAM
without losing speed. This combo uses Dual memory channel while the
754 design uses single memory channel and thus when you use the 3rd
DIMM slot the memory slows down.

When you say slows down, what are you referring to: clock speed or timings?

 

[Derek Baker]

My new athlon XP system just got fried..

I have some questions. I've heard the Athlon64 3200+ can beat a P4
3.2 EE in most tests. Is this true, and is there any reason why one
would buy the P4 over the A64 other than bias?

I'd say some tests, rather than most.

You'd buy the P4 if media encoding was your main task.

Also, i'm a bit confused now.. I see A64 3200+ listed at pricewatch..
some are listed as 1600mhz, and some listed at 2.2ghz. is the 1600mhz
just describing the fsb? and do they all run at 2.2ghz?

Older 3200+s run at 2Ghz, newer ones at 2.2Ghz.

Lastly, I see there are some that have 1mb of cache as apposed to 512
for around the same price. Is there a catch that I'm missing?

Not much.

The 2Ghz models have 1Mb, the 2.2Ghz ones have 512Kb. The reduction makes
very little difference to the Athlon 64 - they upped the clock speed anyway
on those models - and it makes the chips cheaper to produce.

 

[Derek Baker]

Mike, your best bet would be to get an AMD XP 3500 939pin cpu, with a
939pin based motherboard. This way you can get up to 4gigs of DDRAM
without losing speed. This combo uses Dual memory channel while the
754 design uses single memory channel and thus when you use the 3rd
DIMM slot the memory slows down.

Assuming he needs that much memory. I've just ordered a Socket 754 K8N Neo
Platinum which 'only' takes 2GB. Its manual states it will run three DIMMs
at DDR400 speeds, the main reason I bought it.

 

[Jason Cothran]

| George Hizinwak wrote:
| > | >> My new athlon XP system just got fried..
| >>
| >> I have some questions. I've heard the Athlon64 3200+ can beat a P4
| >> 3.2 EE in
| >> most tests. Is this true, and is there any reason why one would buy
| >> the P4 over the A64 other than bias?
| >>
| >> Also, i'm a bit confused now.. I see A64 3200+ listed at
| >> pricewatch.. some are listed as 1600mhz, and some listed at 2.2ghz.
| >> is the 1600mhz just describing the fsb? and do they all run at
| >> 2.2ghz?
| >>
| >> Lastly, i see there are some that have 1mb of cache as apposed to
| >> 512 for around the same price. Is there a catch that i'm missing?
| >>
| >> --
| >> www.geocities.com/mtb2k -personal website
| >>
| >>
| >
| > Mike, your best bet would be to get an AMD XP 3500 939pin cpu, with a
| > 939pin based motherboard. This way you can get up to 4gigs of DDRAM
| > without losing speed. This combo uses Dual memory channel while the
| > 754 design uses single memory channel and thus when you use the 3rd
| > DIMM slot the memory slows down.
|
| When you say slows down, what are you referring to: clock speed or
timings?
|

Latency slow downs associated with ECC registered RAM is what is likely
talking about.

 

[Jason Cothran]

| Mike B wrote:
| > My new athlon XP system just got fried..
| >
| > I have some questions. I've heard the Athlon64 3200+ can beat a P4
| > 3.2 EE in most tests. Is this true, and is there any reason why one
| > would buy the P4 over the A64 other than bias?
|
| I'd say some tests, rather than most.

Most is definately more accurate.

|
| You'd buy the P4 if media encoding was your main task.
|

Which is why most is the most accurate. That's pretty much it for the P4
crowd.

| > Also, i'm a bit confused now.. I see A64 3200+ listed at pricewatch..
| > some are listed as 1600mhz, and some listed at 2.2ghz. is the 1600mhz
| > just describing the fsb? and do they all run at 2.2ghz?
|
| Older 3200+s run at 2Ghz, newer ones at 2.2Ghz.

Ignore the 1600MHz. That is the HTT speed.

Clawhammer processors are the 2GHz, 1MB cache ones. Newcastle cores are the
2.2GHz, 512KB cache ones

|
| > Lastly, I see there are some that have 1mb of cache as apposed to 512
| > for around the same price. Is there a catch that I'm missing?
|
| Not much.
|
| The 2Ghz models have 1Mb, the 2.2Ghz ones have 512Kb. The reduction makes
| very little difference to the Athlon 64 - they upped the clock speed
anyway
| on those models - and it makes the chips cheaper to produce.
|

 

[Wes Newell]

My new athlon XP system just got fried..

I have some questions. I've heard the Athlon64 3200+ can beat a P4 3.2 EE in
most tests.
Yes.

Is this true, and is there any reason why one would buy the P4
over the A64 other than bias?

Some apps that make use of HT will outperform the A64. If you use these a
lot, then a P4 might be the way to go.

Also, i'm a bit confused now.. I see A64 3200+ listed at pricewatch..
some are listed as 1600mhz, and some listed at 2.2ghz. is the 1600mhz
just describing the fsb? and do they all run at 2.2ghz?

The 1600MHz is referring to the hypertransport bus.AFAIK, there's two
models of the 3200+, one is 2GHz with 1M cache, the other is 2.2GHz with
512K cache. Get the 2.2 model. It will be faster in most cases.

Lastly, i see there are some that have 1mb of cache as apposed to 512
for around the same price. Is there a catch that i'm missing?

The clock speed of the cpu is faster on the one with less cache.. There's
many reviews of all the models at tech sites. I suggest you review them
for your applications.before buying. Since the bandwidth of the onboard
controller is so much better than the older method, dual channel doesn't
really bring a lot to the table except to get in your pocketbook. Socket
754 is very close to the performance of 939 all other things being equal.
But the max cpu speed slated for 754 is 3700+, so if you want faster than
that now then 939 would be the way to go. Otherwise, save your cash.
Prices are changing daily, so do your homework.

 

[Derek Baker]

Latency slow downs associated with ECC registered RAM is what is
likely talking about.

Or possibly that some motherboards fall back to DDR333 with three DIMMs, for
example the K8N-E Deluxe that I also considered, and rejected for that
reason.

 

[David Efflandt]

My new athlon XP system just got fried..

I have some questions. I've heard the Athlon64 3200+ can beat a P4 3.2 EE in
most tests. Is this true, and is there any reason why one would buy the P4
over the A64 other than bias?

I when I compared my Dell P4 3 GHz Hyperthreaded PC3200 dual channel at
work with my HP Athlon64 3200+ (2GHz) PC3200 single channel at home, the
P4 beat the AMD64 in all Sandra math, multimedia, and memory tests. Yet
3DMark03 results with the same exact video card (FX 5700LE) were faster on
the AMD64. But it is not a fast enough video card to be cpu limited.
Bumping up HTT 10% on the 1g cache AMD64, which bumped cpu to 2.2 GHz, RAM
to 440, PCI bus (everything except AGP), bumped Sandra scores 10%, but had
almost no effect on 3DMark03 or Doom3 timedemo.

So maybe my AMD64's 1200 MHz FSB with 400 MHz interleave RAM is close to
P4's slightly faster math, 800 MHz FSB and dual-channel RAM in certain
tasks. Note that P4 RAM access in this case was only 68% efficient vs.
95% for the AMD64. So 1200/800 = 1.5 AMD vs. 68 * 2 / 95 = 1.43 P4
plus its math advantage which varies by test.

 

[General Schvantzkoph]

My new athlon XP system just got fried..

I have some questions. I've heard the Athlon64 3200+ can beat a P4 3.2 EE in
most tests. Is this true, and is there any reason why one would buy the P4
over the A64 other than bias?

Also, i'm a bit confused now.. I see A64 3200+ listed at pricewatch.. some
are listed as 1600mhz, and some listed at 2.2ghz. is the 1600mhz just
describing the fsb? and do they all run at 2.2ghz?

Lastly, i see there are some that have 1mb of cache as apposed to 512 for
around the same price. Is there a catch that i'm missing?

I've been benchmarking my new A64 3400+ laptop against my 2.66GHz Xeon
server. I've been running NCverilog simulations, Xilinx place and
routes and GCC (Mandrake 10.0 on the Xeon, Mandrake 10.0 AMD64 version on
the laptop). Doing NCverilog the Athlon 64 is nearly twice as fast as the
Xeon (1.95 to be exact). Place and routes are 1.77 times as fast on the
A64 and and GCC is 1.4x faster, i.e. the performance range is between a
3.68GHz and a 5.2GHz Xeon. The laptop is using the 754 pin 1M cache
version of the A64, Opterons and 959 pin A64s should be even faster.

 

[JK]

If you compared the performance of an Athlon 3400+ desktop to your notebook,
you would see that the desktop would probably be a much faster performer.
Some have suggested replacing the notebook hard drive with a 7200 rpm
one to improve performance significantly.

 

[General Schvantzkoph]

If you compared the performance of an Athlon 3400+ desktop to your notebook,
you would see that the desktop would probably be a much faster performer.
Some have suggested replacing the notebook hard drive with a 7200 rpm
one to improve performance significantly.

The hardrive speed matters only if you are doing a lot of disk accesses.
As long as you have enough RAM there are hardly any disk accesses after
you've booted the system because everything ends up in a RAM cache (at
least in Linux, don't know what XP does). My A64 laptop has a slow 4400RPM
disk which does effect it's performance when writing very large files but
it still beats my dual Xeon server even on those benchmarks. I ran NCsim
in two ways, with recordvars off and on. When recordvars is on it saves a
huge (multigigabyte) file, when it's off it just writes a small log file.
The ratio between the A64 and the Xeon drops from 1.96/1 to 1.38/1 because
of the disk bottleneck, but that's still equivalent to a 3.68GHz Xeon.

 

[Jason Cothran]

| If you compared the performance of an Athlon 3400+ desktop to your
notebook,
| you would see that the desktop would probably be a much faster performer.
| Some have suggested replacing the notebook hard drive with a 7200 rpm
| one to improve performance significantly.
|

I just replaced my 4200RPM drive in my notebook with a 7200ROM one
yesterday. HIGHLY advisable to anyone with a 4200RPM Hard drive.