The Official PC building thread -3rd Edition

lol i dont remember ever saying anything but hell i'll take credit for goodness!

 

Actually I think I do remember you commenting on it, but I'm not sure where or what was said.

 

Sam,
I think the conversation was to do with the way different places do their test reviews, and I remember Shaff asking why not just slow the faster CPUs down to the same speed as a slower one to compare them. I remember answering at the time that I didn't think that it would work that way, but I've found out since that I was wrong, as it seems that it's an excellent way to compare them because it all comes down to the differences in the chip's architecture, rather than trying to make any sense out of trying to compare them at different CPU and Memory speeds. If everything is set the same, then the results should be much easier to see and compare.

Best Regards,
Russ

 

Well, like for like CPUs you can. Fortunately the X4 620 and 940, other than having different amounts of cache, are very similar. You couldn't really do it for fixed multiplier CPUs like Core 2 Quads as the memory speeds wouldn't match up properly.

 

Sam,
No problem, just lower the memory speed to the 1066 speed of say a 620 or 630. That's exactly what Tom's did in the article I posted. They lowered the clock speed of the CPU to the same 2.6GHz of the 620 and the DDR3 memory was run at 1066 for both tests. Bear in mind that the 620 and 630 do not have a unlocked multiplier, either. Also, most motherboards these days will let you lower the multiplier, so it shouldn't be that difficult.

Russ

 

I still wouldn't trust it to be honest, I would rather try and extrapolate the clock speeds. The Toms method is ideal for AMDs, but cross-platform, it has limitations.

 

Sam,
Well, it wouldn't be perfect but it would still be better that the way some places do it now. I read a review from some rinky-dink review site on the 620 and 630 the other day, and the guy that did the review made it plain that he did not like AMD at all. You don't have to lie, but rather just choose the software that favors Intel over AMD, or vice-versa. What can you say about the honesty when they tell you right out front that they don't like one brand or the other, and rip the one they don't like all over the place. LOL!!

I got all three computers finished early this morning., and Russell is thrilled with the Athlon x2 7750. It makes a huge difference in the games he plays. I was pretty impressed with the Foxconn MB I bought to put in Russell's. For the money and the way it runs, it's impressive. Pretty much the same MB he had with the original eMachine, only this one will overclock and can handle 95 watt CPUs. The original was limited to 65 watt CPUs. I'm running a single core 3800+ Lima in mine, for the moment. I should be able to order a 630 by the end of the week! The 3800+ runs pretty good though. I wouldn't want to encode with it, but just for playing around on the net and simple things, it's not as bad as I expected.

Russ

 

Sadly, there are far too many sites like that. Tom's Hardware is one of them, which is why I'm often so suspicious of their tests. Anandtech is not, they just have had some rather crazy mistakes left unedited in their tests in the past.

 

Though not in the forum's title, I see this forum is for building machines optimized for video games. Just wanted to say that everyone appears to be following the (still valid) formula of choosing one's software, then operating system best for it, then hardware best for both.

Just wanted to remark that you can build computers optimized for a variety of activities, from photography to security. Affording only one computer, I simply upgraded a 2002 Mac 'Quicksilver' (optimized by Apple for Unix) to run Debian Linux, whose OS I can tune and modify for my use.

The original goal was to build hardware optimized for the most flexible software (for which I chose Debian GNU/Linux), and run as many ISO & other standard formats & protocols (zillions of sockets) as possible. It lets me write scientific software while handling all security for nearby laptops and answering the phone. I'm hoping to teach it to garden.

Just wanted to remind people that computers can also be hand-built, or rebuilt, for one's specific hobby, to extend the life of laptops, or just to learn computing.

 

I wouldn't say that's the case at all - it plays a significant role, but it's far from everything that matters. We do after all, have a separate thread for that exact purpose.

 

I think it's more a case that AfterDawn has an abundance of gamers.

A 2002 'quicksilver' is an old school G4 isn't it? In which case isn't the coding dramatically different? Doesn't this kill any potential for redistribution or is that not the point of your rather ominous sounding 'scientific software'? I just want to learn as I'm not as up-to-date on these such subject matters as I'd like to be and you seem to know your bearings.

 

Keith,

Sorry, didn't mean to sound ominous. Yes, I just checked 'NIST', and their software is principally for Unix, and the binaries will run under Linux (with X).

Long before PCs, scientists distributed the source code of their applications. (The ACM did as well.) In the 1970s, Unix became the favorite programming platform, and has remained so. 'Old School' scientists believe that science, by definition, is to be offered freely to as many people as want it.

Because Linux is free, has replaced Unix, and runs on many platforms, scientists make available either the f77 (Fortran) code, try and write portable C or C++, or distribute binaries that will run on any flavor of *ix.

Yes, the 2002 machine is old; but MacOSX 1.0 was essentially BSD Unix. It has a 64-bit, 133 MHz bus, 1000 Mbit/sec ethernet port, and everything else can be upgraded. (The processor is on its own board, so it can be doubled & caches included.) It's optimized not for speed, but tuned to run well its virtual memory and all my old gizmos (ports & protocols of all ages), which will be needed if it's ever to garden.

However, it's valuable to read here about processors, power supplies, and fans.

 

Creaky,
Thanks for the link for Auslogics. Fastest disk defrager I've ever used. I just downloaded and installed 3.0.2.40, but I haven't figured out how to defrag all the disks at the same time yet. How do you set it to do that?

Best Regards,
Russ

 

I've not tried the new version, but you could queue them in the old version by ticking multiple drives along the left column, then starting the defrag, perhaps it's a similar method?

 

Analysing drives doesn't take long in AusLogics though, thankfully.

 

My one criticism of it, however, is that it does not bother compressing files very well, which means there are lots of gaps left in the structure between the files. While this shouldn't have any direct impact on performance, it means there's less space to work with bigger files, so you're more likely to run into 'free space not found' problems.

 

I more wasn't sure if there would be differences between the PowerPC coding and x86/x64 or do those sorts of problems only occur at higher levels?

As for defraggers I've used the severely basic Defraggler for aaggggeeessss and it does the job. It might do anything dramatically special or anything but from my understanding it doesn't need to does it?