My review of a newly built Athlon MP 2000+ system

[mrtristan]

I finished building and setting up my new computer. Here's a short review. Feel free to ask any questions you might have, although I've only had it running for two days. It's much too soon to say how it will ultimately work out, but the preliminary results seem to be okay.

First off, the specs.

2x Athlon MP 2000+ (1.67Ghz)
Asus A7266-D Motherboard
2x 512 Crucial 2100 DDR ECC
40Gig Maxtor 7,200RPM ATA100
120Gig IBM 7,200RPM ATA100
ATI AIW Radeon
Creative Soundblaster Live! 5.1

There's some other peripherals, but they don't have any bearing on capturing or encoding.

Onboard sound

Initially, I ran the system using the onboard C-Media 6 channel sound chip. It was okay, but I preferred my Soundblaster. Technically, there was nothing wrong with the onboard audio. The Soundblaster does provide higher quality audio, though.

Stability

Probably what I like most. I upgraded from an Abit KT7A, which is based on the VIA KT133A chipset. This board gave me so many headaches, it's not even funny. If there is one good thing that came of them, I learned a lot about computer components and software.

As for the Asus board, it's rock solid. I can't really say for sure, as I've only had it running for two days, but OS and driver installations went without a hitch. The only problem that I encountered was with ATI MMC. I had problems with my old board, but I gave it a try anyway.

Speed and performance

Feel-wise, the OS seems to run like a Mac. I can't really describe it, but it's just smoother. Again, it's hard to make a judgement, because I've only had it up and running for two days. However, I'm expecting that things will continue to run smoothly, since I haven't encountered any hitches as of yet.

As for the actual speed of the system, I'm a little dissapointed. To tell the truth, I was hoping for blazing fast, although I knew to expect otherwise. Even still, my realisted expectations of a dual CPU machine were a little higher than what it turned out to be.

My testing ran as follows:

Captured 30 second clip at 640X480, using huffyuv and noise reduction enabled, in Virtualdub.

Opened the clip in Virtualdub's editing mode and ran the de-interlace filter and frameserved to Tmpgenc.

My Tmpgenc settings were mpeg2 at 480X480, 2 pass max 5000, min 300, avg 3000. Or something. I'm too lazy to open it up and dig out the exact settings.

Running with no filters and 8 bit precision at high quality, it took roughly three times realtime. Again, I can't recall the exact figure and I'm too lazy to do a test right now.

I'm getting tired and this review is going downhill fast, as my heavy eyelids compell me to hit the bed. I'll add more as I go along and answer any questions along the way.

As of yet, I'm still not sure if I would go with a dual AMD system again, leaning towards the "not" side.

Cooling

I'm glad my case came with 4 fans. Each cpu runs at about 48°C at idle. Pretty damn hot, considering my overclocked Thunderbird used to run at 33°C, even under full load. I went with the retail box, in order to save some hassle and money with third party fans, but it looks like I'm going to have to upgrade. I did some research and 50°C is normal, but I just don't feel comfortable with them running that hot.

[jeex]

can you really run tmpgenc twice and have it encode one movie on each proccessor at full proccessor usage on both proccessors?

[zzyzzx]

You need to update your "Computer Details" section.

[mrtristan]

Okay, I tried it out. I posted an edit last night, but it didn't go through. I ran two separate clips and tmpgenc two times. Each cpu ran at close to 100%, dropping every few seconds. The encoding times actually took longer, though.

The 30 second clip that I started took 2:22, while the first clip took 4:17.

The real benefit comes about when I frameserve with virtualdub and run tmpgenc. The combined cpu usage runs from 70%-90%, but is a lot faster than running the native deinterlace filter in tmpgenc.

can you really run tmpgenc twice and have it encode one movie on each proccessor at full proccessor usage on both proccessors?

[rhegedus]

My P4 2.26 oc to 2.53 runs at 48 degrees after 3 hours of constant 100% load so 50 degrees at idle sounds a bit too hot but, then maybe that's just Athlons for you. I'm curious to know what temp they're at under constant 100% load.

From having built a PC myself, I discovered that airflow within the case is vitally important. Unfortunately, many cases have exhause vents just a few cm from the most lateral PCI slot, so if this is used it will disrupt airflow through the case. I also try and leave a drive bay's worth of space between hard drives to prevent pockets of hot air building up between them.

What sort of a case do you have and how are the fans arranged?

Regards,

Rob

[mrtristan]

Rob

It's a Lian-Li PC-60 case. There are two intake fans and two exhaust fans, for a total of four. One exhaust fan is in the standard location, while the other one is in the center of the top panel. After doing some research, my chips are running within the norm. I tried to tie the cables out of the path of the airflow, so that shouldn't be a problem. When I have some time, I'm going to strip the thermal pads off of the cpus and fans and switch to thermal grease. I didn't do so initially, because I don't plan on overclocking these chips. I assumed that they would run cooler, since the Palomino core athlons run at lower voltage. The whole dual cpu thing must change some of the variables, since they run so damn hot.

The strange thing is, they don't seem to vary much under load and idle. I guess these chips are just inherently hot.

[Faustus]

Yeah AMDs are pretty hot chips. BTW on the dual encodes thing... when you run dual CPU you lose some of the performance on the chips, mostly on the 2nd chip. Thats why neither ran as fast as you expected like that. I think the rule of thumb that was told to me long ago was 60% out of a second CPU is good. (but thats been many moons and therefore tech changes, and I have slept since then)

I'd be interested to know how fast you can do DVD2DVDR CCE encodes with it... any way for you to test?

[mrtristan]

I would run some CCE tests, if I had it. I'm not much of a pirate (arr!), and I can't afford the $2,000 for the version that I want.

What's a DVD2DVDR encode, by the way? A rip off of a DVD and encode back to DVD? I don't have a DVD burner yet, so I'm pretty unfamiliar with anthing related to DVD ripping, burning, or encoding.

I did do a short test with tmpenc, encoding a broadcast capture at 640X480 to DVD format and it still took long enough for me to cancel the encoding process.

If anyone else wants me to run some tests, feel free to ask. I don't mind at all.

[Faustus]

DVD2DVDR is a automation untility that rips a DVD and then encodes it using CCE so that you can do the rest from there... without CCE the point is kinda moot however...

[magomago]

Um...do TMPGEnc and VirtualDub support Multiprocessors? If not you aren't getting any use out of it...

I dunno though...that sounds awsome.. If they are the t-bred A or B cores...OVERCLOCK IT!

[Faustus]

I know TMPGenc and CCE do, not sure about the other.

[mrtristan]

Virtualdub doesn't support multiple cpus, as far as I know.

The chips that I have are based on the Palomino core. They can be overclocked, but they're already running really hot. This system is fast enough as it is, so I don't want to add any extra stress to it.

Um...do TMPGEnc and VirtualDub support Multiprocessors? If not you aren't getting any use out of it...

I dunno though...that sounds awsome.. If they are the t-bred A or B cores...OVERCLOCK IT!