My first build - AMD budget box - Advise please

  • Thread starter Thread starter Cory
  • Start date Start date
I did some digging at the AMD site and you may have been mistaken regarding
the bus speed. It says that the Athlon XP processor has a 400 FSB (and the
Shuttle AN35N supports 400). So the PC3200 would be the best match after
all wouldn't it?

Cory

I've never heard of a 2500+ that has a 400FSB. It could exist, I guess.
I'd double-check those specs. Or, just by the PC3200 RAM. At 333 or 400,
it will work OK. -Dave
 
Okay, here's where my knowledge is admittedly weak. If I use PC3200 with
the Shuttle AN35N, are you saying that it won't make any difference if I run
dual-channel or not? Could you point me to a source that would help me
separate the hype from the facts? Thanks.

Cory
5% gain max. I challenge any normal person to tell if it is really on
dual channel or not. Memory boost is in the FSB speed that's why I say
crank it to 400. That is not so crazy since many get it to 500.
 
Dave said:
I've never heard of a 2500+ that has a 400FSB. It could exist, I
guess. I'd double-check those specs. Or, just by the PC3200 RAM. At
333 or 400, it will work OK. -Dave

My apologies Dave, you were right all along. I didn't look closely at the
web page I was reading at AMD or I would have noticed they were referring to
the XP 3200 which has a 400 FSB. The XP 2500 is indeed 333. As for Ars
Technica recommending PC3200 for that processor, I now gather that it was
indeed for the benefit of those who wish to overclock, just as you
suggested. I appreciate the feedback you provided on this issue and now
feel comfortable with going with PC2700. Thanks.

Cory
 
Though I've never thought I would actually submit to overclocking,
running the 2500+ at 3200+ in my new system under construction (see
above post) by running 400 FSB (actually 200 mhz--right?) is my plan.

I don't know how representative the reviews at Newegg are--I know they
do some editing there; but scads of guys are doing this apparently
with great success and in many case with stock cooling units. So I
figure why not? For this reason I bought Crucial PC3200 ram, 2 - 256
sticks.
I selected Crucial because of superior experience with Crucial. Maybe
Corsair XMS would have been a better choice, but the mission statement
for this new machine is still one of general use.

I'm still wondering if I should have bought a Giga mobo instead of
the A7N8X-E Deluxe. We shall see.

Fedex delivered my Coolermaster Praetorian Aluminum case and it is
very impressive. Maybe not conservative enough for me--but that
removable mobo tray is the bomb. Nice workmanship.
Dave
 
My apologies Dave, you were right all along. I didn't look closely at the
web page I was reading at AMD or I would have noticed they were referring to
the XP 3200 which has a 400 FSB. The XP 2500 is indeed 333. As for Ars
Technica recommending PC3200 for that processor, I now gather that it was
indeed for the benefit of those who wish to overclock, just as you
suggested. I appreciate the feedback you provided on this issue and now
feel comfortable with going with PC2700. Thanks.

Cory

No problem. -Dave
 
Scapaflow said:
Though I've never thought I would actually submit to overclocking,
running the 2500+ at 3200+ in my new system under construction (see
above post) by running 400 FSB (actually 200 mhz--right?) is my plan.

I don't know how representative the reviews at Newegg are--I know they
do some editing there; but scads of guys are doing this apparently
with great success and in many case with stock cooling units. So I
figure why not? For this reason I bought Crucial PC3200 ram, 2 - 256
sticks.
I selected Crucial because of superior experience with Crucial. Maybe
Corsair XMS would have been a better choice, but the mission statement
for this new machine is still one of general use.

I'm still wondering if I should have bought a Giga mobo instead of
the A7N8X-E Deluxe. We shall see.

Fedex delivered my Coolermaster Praetorian Aluminum case and it is
very impressive. Maybe not conservative enough for me--but that
removable mobo tray is the bomb. Nice workmanship.
Dave

I have no experience with overclocking and know very little about it, except
that it's controversial and people here seem to be in one camp or the other.
I'd be interested in any opinions on the pros and cons. I'm speculating
that there might be a heat penalty, which would concern me.

Cory
 
I have no experience with overclocking and know very little about it,
except
that it's controversial and people here seem to be in one camp or the other.
I'd be interested in any opinions on the pros and cons. I'm speculating
that there might be a heat penalty, which would concern me.

Cory

Overclocking is quite simply running PC components faster than they are
designed to be run. At best, an overclocked system is less stable and will
not last as long as a computer that is run at speeds it is designed to be
run at. How long would you expect your car to last if your engine was
constantly pushed past redline?

Now you will hear from gazillions of people who successfully overclock and
claim that their systems are stable, no heat problems, etc. Just whatever
you do, don't buy a used computer from them. :) Overclocking can be done
quite safely. So can jumping off a cliff. I don't advise you try either,
as a bit of luck is required to keep both activities safe. -Dave
 
Okay, here's where my knowledge is admittedly weak. If I use PC3200 with
the Shuttle AN35N, are you saying that it won't make any difference if I run
dual-channel or not? Could you point me to a source that would help me
separate the hype from the facts? Thanks.

Cory, dual channel does give sightly better performance.
It also gives better performance under a number of circumstances.
Primarily if you're able to step up the FSB.
KT600 performs well on singlechannel, but nForce2 is still slightly,
slightly better.

Then there's the question of whether it's significant or not. We're
counting % on the fingers of a single hand here (or single finger).
You'll have to be the judge of that. My first AthlonXP box is nForce,
but today I look a little bit more on money and less at each %
performance. I'd probably buy KT600 today. But I'd feel more
comfortable recommending nForce. nForce2 is a very good chipset. Very
highly regarded.

Here's what I do, (which unfortunately seem to go against most advice
here :-)). I buy one class better ddram than I need. I buy fairly
cheap (Samsung, Kingston value, TwinMOS,.. (if it works, it works)). I
check it works in the box at the rated speed, then I downrate it to
run in sync with FSB.
This means I've spent my money on ram, that will be a little bit more
flexible, for whatever purpose. The obvious, is that a faster cpu
finds it's way into the machine. Other are, moving the memory to
another machine, or 'borrowing' it for debugging purposes, etc.

(I've used exclusively 'cheap' ram for so long it's proof of getting
old, and while I've had modules that didn't work in a particular
machine, their replacement did, and I think overall it's been good
experience. Of course, if you buy mail order, it's a bit more hassle
to switch modules.)

I do another thing too. I buy as _FEW_ sticks as possible! Memory
performance will downgrade much more substantially, than any
difference this thread has discussed sofar, if you have more than two
sticks, or if you have nonidentical sticks. So I'd like to recommend
you to get by with a single 512MB module, (and single channel mode and
slightly lower performance) until you need 1GB ram.
But also again, the bigger modules are more flexible and thus have a
longer useful lifespan.

ancra
 
Cory, dual channel does give sightly better performance.
It also gives better performance under a number of circumstances.
Primarily if you're able to step up the FSB.
KT600 performs well on singlechannel, but nForce2 is still slightly,
slightly better.

Then there's the question of whether it's significant or not. We're
counting % on the fingers of a single hand here (or single finger).
You'll have to be the judge of that. My first AthlonXP box is nForce,
but today I look a little bit more on money and less at each %
performance. I'd probably buy KT600 today. But I'd feel more
comfortable recommending nForce. nForce2 is a very good chipset. Very
highly regarded.

Here's what I do, (which unfortunately seem to go against most advice
here :-)). I buy one class better ddram than I need. I buy fairly
cheap (Samsung, Kingston value, TwinMOS,.. (if it works, it works)). I
check it works in the box at the rated speed, then I downrate it to
run in sync with FSB.
This means I've spent my money on ram, that will be a little bit more
flexible, for whatever purpose. The obvious, is that a faster cpu
finds it's way into the machine. Other are, moving the memory to
another machine, or 'borrowing' it for debugging purposes, etc.

(I've used exclusively 'cheap' ram for so long it's proof of getting
old, and while I've had modules that didn't work in a particular
machine, their replacement did, and I think overall it's been good
experience. Of course, if you buy mail order, it's a bit more hassle
to switch modules.)

I do another thing too. I buy as _FEW_ sticks as possible! Memory
performance will downgrade much more substantially, than any
difference this thread has discussed sofar, if you have more than two
sticks, or if you have nonidentical sticks. So I'd like to recommend
you to get by with a single 512MB module, (and single channel mode and
slightly lower performance) until you need 1GB ram.
But also again, the bigger modules are more flexible and thus have a
longer useful lifespan.

ancra

I don't understand this statement:
"I check it works in the box at the rated speed, then I downrate it to run
in sync with FSB."
What do you mean by 'downrate' here. Is this something to do with BIOS
settings? And are you saying that there's no performance issue with using
RAM rated at a different frequency than the FSB as has been suggested in
this thread? Maybe I'm misunderstanding what you're saying.

Cory
 
Cory said:
I'm comfortable working inside a computer, but when it comes to building,
component selection makes my head spin. I've been looking around for a
reputable website that would help a new builder run this gauntlet. I found
what appears to be a good one at Ars Technica and I'm planning on putting
together a system based on their 'Budget Box' shown here:
http://arstechnica.com/guide/system/budget.html

Here's my current component list:

Motherboard: Shuttle AN35N ($55.12)
Processor: AMD Athlon XP2500 OEM ($80)
Fan: SVC GC68 ($5.50)
RAM: Two 256MB PC3200 DDR ($105.98)
Video: ATI Radeon 9100 128MB ($72)
Hard Drive: Seagate Barracuda 7200.7 80GB ($67)
CD-RW: Lite-On LTR-52327S ($40)
Case: Antec SLK3700BQE ($77)
Monitor: NEC FE991SB ($211.94)
Floppy: Teac 1.44MB ($6.95)
Operating System: Microsoft Windows XP Home ($85)

Total Price: $806.49

Very good setup overall.

Lately I've gotten three 256MB PC2100 free after rebates at OfficeMax.
Also got a good WD 80G drive there for $20 AR.
Look there for rebates on 19" monitors.

The Biostar M7NCD Pro can be substituted for the Shuttle mobo. I have
found that the Biostar is not at all picky about RAM sticks.
 
I don't understand this statement:
"I check it works in the box at the rated speed, then I downrate it to run
in sync with FSB."
What do you mean by 'downrate' here. Is this something to do with BIOS
settings? And are you saying that there's no performance issue with using
RAM rated at a different frequency than the FSB as has been suggested in
this thread? Maybe I'm misunderstanding what you're saying.

It's nothing complicated. Yes, I fiddle the memory speeds in bios.
Like: I need PC2100 (DDR266).
I buy PC2700 (DDR333).
I then check that it really works at DDR333. This is just me being the
customer. If it doesn't, I intend to return it and get another one. I
just wan't to make sure it's alright.
Then I set down memory frequency to 133MHz and DDR266, to match the
FSB speed of 266MHz. I do this because there doesn't seem to be much
reason to run memory faster than the 'bottleneck' FSB. And everybody
says running in sync is worth it.
I then imagine I have very stable ram. Note, I don't know for certain
that this is actually true! That's just how I figure it for now, until
I get better knowledge. ;-)
But as I said stability isn't the real issue for me. The increased
flexibility of use is. As I have several systems, and tinker with them
in various ways.


I don't know what 'they' mean by lower memory performance. They could
simply mean that running async DDR400 on 333FSB is slower than sync at
333. They could also mean that PC3200 is slower at DDR333 than PC2700
at DDR333. I don't know if it is, but I've always assumed, and still
assume, it's not slower.

If it is, possible reasons could be that they assume auto timings from
SPD could generally be more conservative on the faster ram than
slower. But those timings differ for different modules in the same
speed rating as well. I don't bother, except for buying CL 2.5.
Another possible reason could be that refresh is different and
interferes. I don't know. I hope we can get some clarification on
this. Anyway, I'm not worrying about it.

ancra
 
Back
Top