Athlon64: Facts, myths, and experiences...

The forum for non-component-related silent pc discussions.

Moderators: NeilBlanchard, Ralf Hutter, sthayashi, Lawrence Lee

Post Reply
bcassell
Patron of SPCR
Posts: 70
Joined: Tue Mar 23, 2004 1:01 pm
Location: San Jose, CA, USA

Athlon64: Facts, myths, and experiences...

Post by bcassell » Thu May 27, 2004 10:03 am

Hi. As a happy owner of an A64 machine, I've noticed a fair number of misconceptions about the A64 on these boards. This is understandable since it's such a new processor, and not many people have experience with it. Now, I don't claim to know everything about the A64--far from it--but I can share what I know, and what my experiences have been.

First, some general information about the Athlon64:

64 bit-ness
A lot of people seem to hold the opinion that they should wait on purchasing an Athlon64 until the 64-bit software world is ready. After all, that's the whole point of the cpu, right? Well, to be perfectly clear: The Athlon64 supports the 32-bit X86 instruction set 100%. It can and will run any code that your P4 will run. It ALSO supports AMD64/X86-64/IA32-E/whatever-the-hell-else-its-called. Yes, it's true that (in the windows world at least) 64-bit software is not here yet. We don't even have an operating system. But it doesn't matter. The A64 needs to be judged on it's 32bit performance. The fact that it supports x86-64 is simply a (possible) future benifit. Right now, though, it's still one of the best 32bit cpu's you can buy. Which leads me to my next point...

Performance
Regardless of any real or imaginary gains from 64-bit code, when running 32bit code, right now, the A64 is FAST. I'm sure you can find benchmarks all over the net (I like http://www.anandtech.com myself) that will tell you just that. Aside from a few benchmarks (mainly video encoding I beleive) the A64 simply trounces a similarly priced P4.

Personally, I use a P4C 2.8 at work, and an A64 3000+ at home. The system specs are VERY similar (1gb of the same ram, 7200rpm 8mb cache hard drive, etc). I can tell you that, even though the P4 has hyperthreading, the A64 FEELS faster. Things are snappier. Even when multi-tasking the A64 feels faster. In fact, I'd go so far as to say that, when multitasking (say, for instance, web surfing/etc. while compiling), the A64 becomes even more responsive in comparison with the P4. My guess is that in these situations the integrated memory controller makes the biggest difference.

On more thing, though you can get many many benchmarks on many different websites for many different applications, one benchmark I almost never see is code compilation. To give you guys a quick example: the exact same code that takes 16 minutes to compile on my p4 2.8 at work, takes 7:45 to compile on my A64 3000+ at home. Yes, it's over twice as fast.

Cool'n'Quiet
This seems to be one of the most talked about, and most misunderstood features of the Athlon64. Simply put, all this feature does is automatically underclock/undervolt your cpu when it is not being heavily used. I'm sure there are many threads already on these boards that go into much more detail about this feature. If you want a lot more information, and my personal experiences with it, just PM me.

What I see asked a lot is some form of the question "so will the A64 run cooler with Cool'n'Quiet enabled?" Well, honestly, that depends on how you use your computer. However, I think CnQ should NOT be taken into account when considering the heat output of the A64. "But why not?" you may ask. At some point, we all run our computers under full load. If you don't, then why the hell did you buy that nice shiny new cpu? And when running under full load, CnQ does absolutely nothing to affect heat output. So, this brings me to my next point...

Heat/Power
Let me start by saying this: The athlon64 is a very cool cpu when compared with anything else even close to it's performance. I often see people quoting the listed 89W max power output, and claiming that the A64 will be no cooler than the P4. Well, I don't know where AMD gets their power numbers from, but it's obviously not the same place as Intel. I can think of two comparisons I've read recently: a post by MikeC and a cpu heat comparison at anandtech. Both of these taken together would indicate that the A64 is putting out SIGNIFICANTLY less heat than any P4.

Another thing to take into account is undervolting capability. I've dealt with two A64 3000+ cpus so far. The first one was from a very early batch--I got it shortly after the 3000+ was released. It would run completely stable at 1.35 volts. The second one I got about a month ago and it's the cpu I'm currently running. It runs completely stable at 1.3 volts, and might go lower, I just haven't tried. At 1.3 volts, with a panaflo-modded Zalman 7000 AlCu @ 5v (very little airflow), I almost never see the cpu go over 40C. And that's running folding@home 24/7. If I turn the fan up to 11v, the temp won't go above 35C. This cpu is COOL.

Well, there's a ton more that I'd like to write, but I've got to get back to work, and I think I covered the basics. Sorry for the long post, but I really just wanted to clear up some misconceptions and share my experiences with the Athlon64.

Bryan

tragus
*Lifetime Patron*
Posts: 356
Joined: Fri Apr 18, 2003 11:19 am
Location: Baltimore, MD

Post by tragus » Thu May 27, 2004 10:16 am

Thank you very much sharing your experiences. I'm very pleased with your positive reaction. I'll admit to have my share of misconceptions, though I also admit to remaining skeptical about the 100% backward compatibility (at least at the fringes today's software/drviers that can be touchy enough). I'm hoping to be convinced for my specialized work needs. However, if I were to buy a personal machine, I'd seriously consider an AMD 64 system.

MikeC
Site Admin
Posts: 12285
Joined: Sun Aug 11, 2002 3:26 pm
Location: Vancouver, BC, Canada
Contact:

Post by MikeC » Thu May 27, 2004 10:30 am

bcassell --

Clear, to the point, thorough: You hit all the nails cleanly on the heads. Thank you. :)

PS -- would you be interested in writing for SPCR? 8) :lol:

BTW, my A64-3200 system right next to my main P4-2.8C rig. Win XP Pro on both. No contest: The A64 runs faster & cooler. I don't mean benchmarks, I mean using the full range of apps I use daily -- Photoshop, Dreamweaver, Acrobat, Adobe InDesign, a bunch of web design tools... I've been gradually migrating to the A64 -- to turn it from backup machine to main machine. I have not seen a single instance of (32-bit) software not running properly on the A64.

bcassell
Patron of SPCR
Posts: 70
Joined: Tue Mar 23, 2004 1:01 pm
Location: San Jose, CA, USA

Post by bcassell » Thu May 27, 2004 10:42 am

tragus wrote:... I'll admit to have my share of misconceptions, though I also admit to remaining skeptical about the 100% backward compatibility (at least at the fringes today's software/drviers that can be touchy enough) ...
Well, personally I don't really think of it as "backwards compatibility." Let me put it this way: When Intel added SSE to their cpus, did you worry about it breaking non-SSE applications? When they added SSE2, did you worry about it breaking SSE applications? I know I didn't. AMD has now added a new feature: x86-64. I may have my doubts about how well that new feature will work, but I'm not worried about the addition of that new feature breaking x86 support.
MikeC wrote:Clear, to the point, thorough: You hit all the nails cleanly on the heads. Thank you.

PS -- would you be interested in writing for SPCR?
Wow, I don't know if you're serious or not but I'm flattered none the less. If I had realized my writing was going to be seriously critiqued I would have spent more time on that post =) (I'm at work, so I just wrote it as fast as possible). Anyway, if you really were serious then please send me a PM, though I don't know what more I could bring to this already outstanding site =).

Bryan

tragus
*Lifetime Patron*
Posts: 356
Joined: Fri Apr 18, 2003 11:19 am
Location: Baltimore, MD

Post by tragus » Thu May 27, 2004 3:31 pm

OK, I'm starting to be convinced. I'm holding off for a few days on a purchase decision that I was about to make today.

I've done a bit of searching/reading and am impressed with the write-ups I've seen. As usual, benchmarks are a bit like Disraeli's statistics. I want to know how a system will run with my applications. I've tried in vain to find a good comparison/discussion of Intel's HT (multiple logical processors) versus AMD HT (multiple data pathways to chipsets, if I understand).

My mission-critical lab machines, in fact, run two major independent applications at the same time with one controlling and communicating to the other (MATLAB and a data acquisition package called Neuralynx/Cheetah). I've visually monitored CPU usage and don't have a clear picture on how the work is divided. At times, one process clearly takes up all of one logical CPU , while the other takes about a quarter to a half. Other time the split is completely uneven or neither process is chewing many cycles (except for folding, of course).

From my understanding, the "single" AMD could devote all of its resources when one application is in heavy processing and should therefore fare better than the Intel HT. If Microsoft's multi-tasking algorithm (e.g., context switching) is smart enough, then there should be a net overall gain even when both applications are going full tilt. Or does the Intel HT really provide an advantage of balancing and evening out the workload (processing thoughput) without going to a true multi-CPU solution.

Any pointers to articles or other good write-ups to help clarify will be greatly appreciated.

bcassell
Patron of SPCR
Posts: 70
Joined: Tue Mar 23, 2004 1:01 pm
Location: San Jose, CA, USA

Post by bcassell » Thu May 27, 2004 5:38 pm

tragus wrote:...I've tried in vain to find a good comparison/discussion of Intel's HT (multiple logical processors) versus AMD HT (multiple data pathways to chipsets, if I understand)....
Don't let the similar names fool you, HyperThreading and HyperTransport are completely unrelated technologies. There are many detailed discussions of these technologies, but in short:

HyperThreading is Intel's name for their SMT (Simultaneous multi-threading) technology. A HyperThreading-enabled CPU exposes itself as 2 logical processors. Internally, Intel has designed their P4's such that certain stages of the cpu can be used at the same time. By exposing 2 logical cpus, the OS can schedule 2 threads to run at once. The cpu then tries to run both those threads at once. Sometimes, both threads will need to access parts of the cpu that cannot be shared, and you will see no speed increase by running them simultaneously (in fact, sometimes you can see a speed loss). Other times, they will be accessing different parts of the cpu that can be run simultaneously, and you will see a speed increase.

HyperTransport is AMD's name for their cpu-connect bus. It's simply the bus that connects the cpu to, well, everything. Athlon64 cpu's will have a HT link to the memory, to the northbridge, and, in multiple processor configurations, to each other.
tragus wrote: My mission-critical lab machines, in fact, run two major independent applications at the same time with one controlling and communicating to the other (MATLAB and a data acquisition package called Neuralynx/Cheetah).
[...]
From my understanding, the "single" AMD could devote all of its resources when one application is in heavy processing and should therefore fare better than the Intel HT. [...]
Ok, there's a whole lot to explain here... a WHOLE lot, but it's really outside of the scope of this post. If you really want more information, PM me.

That said, it is possible that hyperthreading could help in your situation. It is also possible that it could hurt, or neither. But that only matters when comparing a P4 with and without Hyperthreading. What you really want to know (I'm assuming) is: "Which will run my two applications faster, a P4 or an A64?". Unfortunately, the only way to really know is to try it. The differences between the platforms are so huge that it's impossible to say which one would work better for you without knowing all the details about how the applications you are running are written. The only way you're going to find out is to test them.

Gotta go back to work, sorry I don't have time to write more.

Bryan

TheWesson
Posts: 101
Joined: Tue May 25, 2004 9:41 pm

Re: Athlon64: Facts, myths, and experiences...

Post by TheWesson » Fri May 28, 2004 12:28 am

bcassell wrote:
On more thing, though you can get many many benchmarks on many different websites for many different applications, one benchmark I almost never see is code compilation. To give you guys a quick example: the exact same code that takes 16 minutes to compile on my p4 2.8 at work, takes 7:45 to compile on my A64 3000+ at home. Yes, it's over twice as fast.

Bryan
Interesting point.

AMD CPUs will excel vs P4 in tasks that involve a lot of (unpredictable) branching. The long pipeline of the Pentium will kill it on a branch mispredict - the pipeline is dumped and must be refilled.

I suspect compilers do a lot of complicated (unrepetitive) branching.

So if you are a programmer, get AMD.

However, if you work with multimedia, P4's may be a better choice. Anything that has to stream gobs of data thru the CPU ... The A64 has caught up a lot in that respect, however, due to the integrated memory controller.

Anyhow, my 1533 AMD M4 (Tbird) is a faster-compiling machine than my P4 2.4 at work ... so I concur with bcassell.

the wesson

NeilBlanchard
Moderator
Posts: 7681
Joined: Mon Dec 09, 2002 7:11 pm
Location: Maynard, MA, Eaarth
Contact:

32bit is *native* for the Athlon 64

Post by NeilBlanchard » Fri May 28, 2004 5:50 am

Hello:

32bit programs and OS's are perfectly compatible with the Athlon 64 -- remember the 64bit capabilities are *extensions* to the 32bit, so there is absolutely not compatibility issues. Running 32bit is native, and the 64bits will become available when you run a 64bit OS with a 64bit app.

The Athlon 64 kicks some serious booty! And it runs very cool -- I just built a 2800+ system that I have undervolted to 1.35v and it barely gets above 40C while folding -- it is just 11C above the case temp! :P :twisted: 8)

Michael_qrt
Posts: 154
Joined: Thu Oct 17, 2002 5:53 pm
Location: Sydney

Re: Athlon64: Facts, myths, and experiences...

Post by Michael_qrt » Fri May 28, 2004 6:15 am

bcassell wrote:I often see people quoting the listed 89W max power output, and claiming that the A64 will be no cooler than the P4.
I believe that this is the figure set by AMD for the max allowable power dissipation on the S754 platform. That is, it bears no relation to the actual power output of any processors that fit in S754. It is listed as "thermal design power" (TDP) which I have taken to mean "design general thermal solutions to dissipate up to 89W of power".

Anyway, I read somewhere that AMD is specifying a higher TDP for the S939 platform, around 105W if I remember. This is just to allow for any future higher power processors that they might release (or might not).

There is no reason why a A64 with the same core and clock speed should dissipate much more power on S939 than on S754 so this further suggests to me that the figure is more related to the platform rather than the processor itself.

crypt0r
Posts: 61
Joined: Sat Oct 19, 2002 3:57 am

Very true

Post by crypt0r » Fri May 28, 2004 7:00 am

I built an A64 3200+ ASUS K8V Deluxe CNPS7000CU system back in January.

I must say that cool 'n quiet is not much to get excited about. However, cool 'n quiet + speedfan is a dream come true. CnQ declocked the CPU to the tune of 800 or so mhz, and I believe it devolted the CPU, too.

It doesn't take much airflow through a huge copper heatsink to cool a devolted 800mhz CPU!

There isn't a performance penalty with CnQ enabled. It takes virtually no time for the CPU to reach maximum clock when the CPU becomes heavily loaded.

CnQ + Speedfan = dream team...now for that PSU...

CharlieChan
Patron of SPCR
Posts: 198
Joined: Sun Jul 13, 2003 2:57 am
Location: East Anglia, UK

Post by CharlieChan » Fri May 28, 2004 7:08 am

Does anyone know the power drawn at the socket for a A64 3000? This will probably give some idea of power requirement of the A64. My P4 3.0C with 1 seagate barracuda IV 80G, 1 seagate barracuda 36ES2 18G, Ti500 and a SS300FT draws 155W when folding.

MikeC
Site Admin
Posts: 12285
Joined: Sun Aug 11, 2002 3:26 pm
Location: Vancouver, BC, Canada
Contact:

Post by MikeC » Fri May 28, 2004 8:03 am

CharlieChan wrote:Does anyone know the power drawn at the socket for a A64 3000? This will probably give some idea of power requirement of the A64. My P4 3.0C with 1 seagate barracuda IV 80G, 1 seagate barracuda 36ES2 18G, Ti500 and a SS300FT draws 155W when folding.
I think more relevant than socket power draw is system power -- like what you've specified for yours. It tells you the total heat in the system. I presume that is AC watts? This post will give you lots of details about my A64-3200 setup:

http://forums.silentpcreview.com/viewto ... 74&start=9

Sardaan
Posts: 42
Joined: Mon Apr 12, 2004 6:12 pm

Post by Sardaan » Fri May 28, 2004 9:44 am

I decided about a month ago that my next system will be an AMD 64 (maybe an fx if extra money starts burning a hole in my pocket). At this time I am awaiting the arrival of socket 939 and PCI Express for the next generation video cards. AMD processors have always provided more bang for the buck IMHO.

bcassell
Patron of SPCR
Posts: 70
Joined: Tue Mar 23, 2004 1:01 pm
Location: San Jose, CA, USA

Re: Athlon64: Facts, myths, and experiences...

Post by bcassell » Fri May 28, 2004 9:56 am

TheWesson wrote: [...]
AMD CPUs will excel vs P4 in tasks that involve a lot of (unpredictable) branching. The long pipeline of the Pentium will kill it on a branch mispredict - the pipeline is dumped and must be refilled.

I suspect compilers do a lot of complicated (unrepetitive) branching.
[...]
Branch midpredicts definitely play a factor, which is why the AthlonXP is also faster at compiling than a comparable P4. But don't discount the effect of the A64's integrated memory controller. Compilers work over very large datasets--much more than can be fit in cache. I'd imagine that the P4 spends a lot of time with it's pipelines stalled waiting for memory reads. With the memory controller integrated on the cpu, the A64 has a much lower latency for small memory reads. I think this probably is one of the biggest factors in compile performance.
crypt0r wrote: I built an A64 3200+ ASUS K8V Deluxe CNPS7000CU system back in January.

I must say that cool 'n quiet is not much to get excited about. However, cool 'n quiet + speedfan is a dream come true. CnQ declocked the CPU to the tune of 800 or so mhz, and I believe it devolted the CPU, too.
Yes, at it's lowest speed, with the current (C0) stepping A64's it will underclock to 800mhz and undervolt to 1.3 volts.

In the first incarnation of my A64 system I had the same setup as you, except with a 3000+. I used CnQ + Speedfan (this was before I started folding) and it drove me nuts!!!. What happened was that speedfan would actually turn off my cpu fan when my computer was idle. When I started using my computer, it would wait till the cpu was about 55C, then start ramping up the fan... and keep ramping it up... eventually running it at the full 12 volts (which is not quiet). As the temps dropped, it would slow down the fan... slower... slower... then off. Then the whole cycle would repeat again. Personally I found the constantly changing fan speed much more annoying that a consistent, low-speed fan.

Bryan

NightWatcher
Posts: 8
Joined: Fri May 28, 2004 10:00 am

AMD 64

Post by NightWatcher » Fri May 28, 2004 10:31 am

Greetings all,

I am new to your forums and this is my first post, although I have been reading through them for some time. I couldn't help but take note of the positive experience that was posted with regard to the AMD 64. I have no doubt that the performance of the AMD 64 at this time is very good; however, I caution those of you who have not bought one yet for one reason.

I have been considering an upgrade from an emachine AMD 2200--please don't laugh--I have built much better systems for family members and now it is my turn. In reviewing upgrade options, you should keep in mind that yes, the socket 754 AMD 64 will run 32 bit apps somewhat faster than a P4--and keep in mind it is doing this with SINGLE CHANNEL memory. But there is a caveat--in a few weeks, AMD will begin launching Socket 939s which will eventually have DUAL CHANNEL memory capability. Now, those of you who go out now and buy socket 754 will be stuck with it--unless you are willing to plunk down more money in a few months to get socket 939, which will also involve buying a new motherboard. This marketing technique by AMD is a kicker, because once the socket 939s come out, the 754s will be--what--AMDs equivalent to the Celeron? Not only that, but the 939 which is what we all really want, is going to be very expensive.

Now, I applaud AMD for bringing 64 bit processing to the desktop, but they are not doing justice to the consumer. I would wait for the socket 939s, the BTX mobos and the DDR2 memory support, but I am going to have to pay around $600 for that 939 when it comes out; or limp along longer with my AMD XP2200 until the price comes down. I'm sure I am not the only one who has seen this coming.

Is there anyone else in this boat? :?

Sardaan
Posts: 42
Joined: Mon Apr 12, 2004 6:12 pm

Post by Sardaan » Fri May 28, 2004 10:37 am

Read my post above, I to am waiting on the 939 mother boards. As for BTX for factor and DDR2 support, I am not holding my breath. From what I have heard DDR2 is Q4 of 2005 and I am not hearing enough about BTX to keep me excited.

bomba
Posts: 320
Joined: Fri Jan 16, 2004 12:36 pm
Location: USA

Post by bomba » Fri May 28, 2004 12:23 pm

Sardaan wrote:Read my post above, I to am waiting on the 939 mother boards. As for BTX for factor and DDR2 support, I am not holding my breath. From what I have heard DDR2 is Q4 of 2005 and I am not hearing enough about BTX to keep me excited.
MOO (My Opinion Only): If you keep waiting for the next big thing to come to market, you'll always be waiting!! And, once you plunk down the cash, you'll likely be paying too much for priviledge of helping the component manufacturers iron out the many bugs that always seem to be present in newly released bleeding edge hardware.

It usually makes better sense to keep a half step back from the bleeding edge. From recent posts on SPCR, the current generation A64 sounds like a winner and the prices seem to be rather decent as well.

bcassell
Patron of SPCR
Posts: 70
Joined: Tue Mar 23, 2004 1:01 pm
Location: San Jose, CA, USA

Re: AMD 64

Post by bcassell » Fri May 28, 2004 12:34 pm

NightWatcher wrote:Greetings all,

I am new to your forums and this is my first post, although I have been reading through them for some time.
Welcome to SPCR!!!
NightWatcher wrote: Now, I applaud AMD for bringing 64 bit processing to the desktop, but they are not doing justice to the consumer. I would wait for the socket 939s, the BTX mobos and the DDR2 memory support, but I am going to have to pay around $600 for that 939 when it comes out; or limp along longer with my AMD XP2200 until the price comes down. I'm sure I am not the only one who has seen this coming.
Certainly the coming of socket 939 is something that should be taken into account. I would ask you this, though (and be honest): How many times have you upgraded your cpu without upgrading your motherboard as well? I can only think of one time I've ever done this--upgrading from a P133 to a P200 on the same motherboard.

That said, you're looking at MAYBE a low single-digit percentage increase in speed in the most memory bandwidth intensive applications when going to dual channel. Just look at benchmarks comparing same-speed A64's to FX's or Opterons. There's practically no difference. Also keep in mind that all 939's will have 512k cache (whereas some 754s have 1mb) and 939s will be more expensive for the same clock speed!

I see the "wait for 939" argument all the time. In fact, I see the "wait for X" argument all the time, where X is whatever the current about-to-come-out technology is. There will always be something to wait for. The fact is that socket 754 A64's are here now, and are great cpus on a performance/watt scale. I agree that AMD has really mucked up this whole socket thing, but I think a lot of it is because of pressure from "enthusiasts" who get too caught up in the most recent buzzwords and don't pay attention to what actually matters--real world performance.

Ok, this rant has gone on long enough, but to summarize:

- Socket 754 represents a great platform for those looking for a high-performance cpu with low heat.
- You most likely will not be able to buy AMD's top of the line cpu in socket 754 format a year from now.
- Socket 939 will offer a longer upgrade path for those looking for drop-in cpu replacements
- Socket 939 will cost MORE money for the same performance as a socket 754 system, and intially will not offer any speed advantages over 754 even with high end processors.

Well, that's my take on things,

Bryan

MikeC
Site Admin
Posts: 12285
Joined: Sun Aug 11, 2002 3:26 pm
Location: Vancouver, BC, Canada
Contact:

Post by MikeC » Fri May 28, 2004 12:46 pm

bomba wrote:MOO (My Opinion Only): If you keep waiting for the next big thing to come to market, you'll always be waiting!! And, once you plunk down the cash, you'll likely be paying too much for priviledge of helping the component manufacturers iron out the many bugs that always seem to be present in newly released bleeding edge hardware.
Yup. MOOT -- My Opinion Only, Too! :lol:

Besides, if it is fast & powerful enough, it is fast & powerful enough! Who cares if the next one has dual mem, really, what do you do with your computer that you need more? I have a P3-1.1G Tualatin system I use from time to time and realistically, I would be perfectly happy with that system for 99% of the stuff I do. The only true reason I need or even want more is because:

1) I know it is available (the usual human greed...)
2) It's hard to run a hardware site without keeping up with new hardware
It usually makes better sense to keep a half step back from the bleeding edge. From recent posts on SPCR, the current generation A64 sounds like a winner and the prices seem to be rather decent as well.
MOOT -- The same price P4 loses to the A64, esp at higher clocks (it seems).

Pricewatch best prices:

A64-3000 -- $205 P4-3.0C -- $193
A64-3200 -- $259 P4-3.2C -- $262
A64-3400 -- $369 P4-3.4 -- $399

---

PS -- agree with bcassell's latest points above, too, esp. summary...

Sardaan
Posts: 42
Joined: Mon Apr 12, 2004 6:12 pm

Post by Sardaan » Fri May 28, 2004 2:31 pm

Good points, I have never upgraded a CPU without upgrading the motherboard also. Only once did I upgrade my CPU/MB where I continued using the same memory.

I won't be upgrading my current setup until late the second half of this year anyway, so I can take the wait and see attitude :-)

Most of my need to upgrade is the same greed (same reason I needed to upgrade all my golf clubs last summer!) that Mike refers to. I also want to tool around with water cooling my next system, being it's way to hot on Orlando :twisted:

Kwiet
Posts: 43
Joined: Fri May 28, 2004 2:56 pm
Location: S. Central Turkey
Contact:

Post by Kwiet » Fri May 28, 2004 3:24 pm

I am not new, just have not logged in and forgot my password! My email address has changed so I decided to start over.

I do what you would call "stealth" upgrades. A gradual process that evolves over a period of three years or so. My case is 2001, half the memory is 2001, PSU is 2003, mobo is 2003, the other half of memory is 2002, fans are 2001/2004, video card cooling system is 2001, Thermalright heatsink is 2002 so you get the idea.

A month or so ago, my latest parts were AMD Athlon 2400 mobile CPU undervolted to 1.40 volts and running at 148x12.5 (max memory will go) at 1.85GHz it is plenty fast enough and runs cool and quiet at 35C idle and 47C loaded with CPU Burnin and 3DMark running at the same time. Threw in a new Radeon 9600 Pro at the same time as those upgrades tend to match.

Since my computer has run three different AMD Athlon CPUs, I like to have an upgrade path. I am generally screwing around with overclocking and the like so if I nuke a CPU 6 months down the line, it was upgrade time anyway!

Next spring, my boys get this box (expect the video card) and I will go with the 939 Athlon64. Thank you for all the excellent information about your experiences with the Athlon64.

crypt0r
Posts: 61
Joined: Sat Oct 19, 2002 3:57 am

Re: Athlon64: Facts, myths, and experiences...

Post by crypt0r » Sat May 29, 2004 1:39 am

bcassell wrote:
TheWesson wrote: [...]
Yes, at it's lowest speed, with the current (C0) stepping A64's it will underclock to 800mhz and undervolt to 1.3 volts.

In the first incarnation of my A64 system I had the same setup as you, except with a 3000+. I used CnQ + Speedfan (this was before I started folding) and it drove me nuts!!!. What happened was that speedfan would actually turn off my cpu fan when my computer was idle. When I started using my computer, it would wait till the cpu was about 55C, then start ramping up the fan... and keep ramping it up... eventually running it at the full 12 volts (which is not quiet). As the temps dropped, it would slow down the fan... slower... slower... then off. Then the whole cycle would repeat again. Personally I found the constantly changing fan speed much more annoying that a consistent, low-speed fan.

Bryan
The system isn't mine, I just built it for someone else...

Your post leads me to believe that you haven't played with Speedfan enough. You CAN set it to a constant low speed if you desire. You can set the range realistically from 25% - 100% speed on a CNPS7K. I believe at 800mhz and 1.3v, the fan speed on the CNPS required to cool it is inaudible. I had fairly good airflow in the case, and I set the ramping to 1% increments. It was/is a spectacular setup.

bcassell
Patron of SPCR
Posts: 70
Joined: Tue Mar 23, 2004 1:01 pm
Location: San Jose, CA, USA

Re: Athlon64: Facts, myths, and experiences...

Post by bcassell » Sat May 29, 2004 10:27 am

crypt0r wrote: Your post leads me to believe that you haven't played with Speedfan enough. You CAN set it to a constant low speed if you desire...
That's entirely possible. I only played with speedfan for a day or two before I started folding 24/7. Now I'm just using a fan controller =)

Bryan

sneaker
Posts: 133
Joined: Tue Dec 24, 2002 8:37 pm
Location: Australia

Re: Athlon64: Facts, myths, and experiences...

Post by sneaker » Sat May 29, 2004 2:00 pm

bcassell wrote:What I see asked a lot is some form of the question "so will the A64 run cooler with Cool'n'Quiet enabled?" Well, honestly, that depends on how you use your computer.
I think this is 100% correct.
However, I think CnQ should NOT be taken into account when considering the heat output of the A64. "But why not?" you may ask. At some point, we all run our computers under full load. If you don't, then why the hell did you buy that nice shiny new cpu? And when running under full load, CnQ does absolutely nothing to affect heat output.
But I think this is a flawed way of looking at it. It basically implies we shouldn't take CnQ into account at all.

You got it right the first time: it depends on how you use your computer. And if you consider the usage patterns of most people, the amount of time spent at full utilisation is relatively small. I suspect that even most enthusiasts spend most of their time on low-utilisation tasks like reading e-mail, browsing the web, creating and editing text documents, and maybe playing music in the background.

And that's when you're actually at the PC. Add the eight hours you may leave their computer running through the night so that you don't have to close all your programs and lose your context in them, or so that you can leave a P2P program running to complete large downloads. Add the hours during the day that you're away from the PC but leave it running for similar reasons.

From a power consumption point of view it's an obvious benefit (if more for the enivironment than your wallet). From a noise perspective, it means that you can achieve an extra degree of quietness most of the time, particularly during those tasks that require you to think. And if you're a young person stuck sleeping in the same room as the PC, it must be a godsend.

I believe a better way of making the point would be to simply to remind people that the Athlon 64 is a high-performance desktop CPU at heart, that CnQ helps little during high-utilisation tasks, and therefore an appropriate thermal solution is required to handle this without becoming too noisy.

I should add that my low-utilisation argument falls to bits if the Athlon 64 throttles to full speed the moment you start playing an Ogg. Based on what I've read, this is not the case. But it's something we need more feedback on--at which point the Athlon 64 is pushed to full speed? 10% utilisation? 30%? 50%? If an MP3 player doesn't bump it up to full speed, does playing a DVD? Would love to hear from youself and other owners on that issue.

JWarrior
Posts: 26
Joined: Tue Feb 18, 2003 4:58 pm

Post by JWarrior » Sun May 30, 2004 2:44 am

Thank you for sharing your experiences on this processor. I am planning on building a new computer before the start of the next school year, and am still debating which processor to go with.

I want to get an Athlon 64, I really do. There is no doubt in my mind that AMD has trumped Intel's top offerings with this chip. Unfortunately, the processor in only half of the story when it comes to building a new system.

The other half is the chipset. This is where I'm leery of taking the AMD plunge again. From the legendary BX on forward to today's 865PE and 875P chipsets, Intel has always paired its processors with solid, stable, platforms with reasonable performance figures. (The RAMBUS boards notwithstanding.)

With AMD, it always seems to be hit and miss. VIA chipsets are always fluky, it always takes two revisions before they put out a chipset that seems to work as advertised (KT133->KT133A, KT266->KT266A, etc.) Even now there's reports that there are conflicts with VIA boards and Sound Blaster products. NVIDIA's NForce2 was a solid chipset for the Athlon XP's, but the NForce3 bits don't seem to be quite as earth shattering. I don't know enough about SIS, but I haven't read anything to make me think that they are the ultimate A64 platform.

For those of you who are running Athlon 64's, what platform did you pair it with. Why? What are your thoughts on the platform? Any stability issues? Any compatibility issues? Anything else that may be of help to a future AMD buyer?

Thanks in advance.

DGK
Posts: 70
Joined: Sun May 18, 2003 3:59 pm

Post by DGK » Sun May 30, 2004 5:53 am

I agree that some of the AMD boards have had some bad reps as far as the chipsets but I have never had a problem with them, most of the people that I have seen have problems are overclockers. I have had 2 Asus boards with VIA chipsets and both were stable.

I currently have a K8V Deluxe/3200+ AMD64 setup with 1 gig of ram and 2 samsung hd's. So far I have been amazed at the chips performance and it runs cool with the fanmate on the zalman cooler set to its lowest possible setting (5V?). As far as Cool&Quiet goes, I think its the best thing sense sliced bread :lol:

The main purpose of this machine is probably 50% web surfing, 25% games and 25% video editing. I know Intel's are supposed to be better for video editing/multimedia but I am happy with the A64's performance in this area.

BTW, I am not a 'fanboy' and am not saying 'amd is great and intel sucks'. My previous computer had a intel chip and I was very happy with it and would buy Intel again in the future. I just think that the A64 is better than anything Intel has out right now in terms of price, performance and heat/power.

MikeC
Site Admin
Posts: 12285
Joined: Sun Aug 11, 2002 3:26 pm
Location: Vancouver, BC, Canada
Contact:

Post by MikeC » Sun May 30, 2004 8:50 am

Not enough exposure to A64 boards to comment generally, but...

The new nVidia 250 chipet board by MSI is getting great reviews. Seems better than the 150 chipset, which is what my AOpen AK89 MAx is based on. Don't know about undervoltability on that board -- the AOpen is fantastic in that regard I have not had problems with this board, tho CnQ is disabled in the BIOS, which is kind of frustrating, as I think I might be able to make the machine a couble dBA quieter with it.

Trip
*Lifetime Patron*
Posts: 2928
Joined: Wed Sep 24, 2003 7:18 pm
Location: SC

Post by Trip » Sun May 30, 2004 11:27 am

bcassell wrote:Yes, at it's lowest speed, with the current (C0) stepping A64's it will underclock to 800mhz and undervolt to 1.3 volts.

In the first incarnation of my A64 system I had the same setup as you, except with a 3000+. I used CnQ + Speedfan (this was before I started folding) and it drove me nuts!!!. What happened was that speedfan would actually turn off my cpu fan when my computer was idle. When I started using my computer, it would wait till the cpu was about 55C, then start ramping up the fan... and keep ramping it up... eventually running it at the full 12 volts (which is not quiet). As the temps dropped, it would slow down the fan... slower... slower... then off. Then the whole cycle would repeat again. Personally I found the constantly changing fan speed much more annoying that a consistent, low-speed fan.
bcassell, the fan would cycle from fast to slow the entire time you were pushing the CPU or would it cycle fast to slow as you went from idle to heavy load?

You could set the fan to run at 10V max rather than 12V max with a fanmate, provided it didn't overheat the CPU, so that max cooling would be less extreme. The fan wouldn't cool the sink off as quickly either.

bcassell
Patron of SPCR
Posts: 70
Joined: Tue Mar 23, 2004 1:01 pm
Location: San Jose, CA, USA

Post by bcassell » Sun May 30, 2004 12:43 pm

Trip wrote: bcassell, the fan would cycle from fast to slow the entire time you were pushing the CPU or would it cycle fast to slow as you went from idle to heavy load?

You could set the fan to run at 10V max rather than 12V max with a fanmate, provided it didn't overheat the CPU, so that max cooling would be less extreme. The fan wouldn't cool the sink off as quickly either.
Yes, under heavy load the fan would constantly cycle from fast to slow to fast to slow, etc. It was really annoying. It doesn't matter now though, I'm running folding 24/7 so I disabled CnQ. Since then I've undervolted to 1.3v and replaced the zalman fan with a panaflo which is running at 5v. My cpu stays nice and cool and my cpu fan is practically inaudible.

Bryan

Post Reply