Intel 6xx Series Reviewed and Benchmarked
sebFlyte writes "It's been a long time coming, but Intel's first 64-bit desktop chip (the 6xx series) is here now, and thanks to ZDNet it has been thoroughly tested. The article has the full specs of the new family, explains the benefits of the changes, and also the results of tests on the new chips to establish perfomance boosts for games, photo manipulation and video work, among other things."
I'm not impressed. All those 32-bit benchmarks to benchmark their 64-bit CPU. Last week Linux Hardware benched the new Pentium against Opteron with real 64-bit apps on a real 64-bit OS.
"Only in their dreams can men truly be free 'twas always thus, and always thus will be."
--Tom Schulman
The only small caveat is that those are, according to you, max power consumption. Typical power consumption may be different. However, the proportion of typical to maximum is most likely comparable between the AMD and Intel chips. Assuming that it is, then the numbers are *still* indicative of heat... :)
Linux IT Consulting and Domino Development in Michigan
I haven't looked into the Intel implementation of x86-64, but I would think using different register sizes would break just about anything.
What you might be referring to is the 4-stage translation table which only allows each process a 48-bit address space. AMD uses this, and they're going to bring it up to the full 64-bit at a later date. I am assuming that despite Intel's implementation being slower, they both are pretty much identical.
Moore's law will have to be fulfilled in new, unique ways.
Moore's Law was never about speed in gigahertz, but rather about transistor counts, so it'll continue on as it always has, since more cores means more transistors.
I wholly agree with you that the examples used are poor, but have you never wanted or needed to get something done whilst something else processor-intensive was happening in the background?
Like playing a game or watching an AVI, MPG or DVD whilst compiling, ripping a audio CD or burning a data CD-R in the background?
"Accept that some days you are the pigeon, and some days you are the statue." - David Brent, Wernham Hogg
If I were you I'd get an older board (i've got an Asus SK8N, works great) and an opteron 1xx. You'll basically have a FX, but for less money and you'll pay less for more L2 cache. The only downside is you need to buy registered ECC ram.
Perhaps it's your choice of OS, or a lack of proper support/configuration for your chipset.
I can quite easily burn a DVD over NFS, while streaming video from the same NFS, chatting in a few apps, browsing, and compiling. Not that I recommend doing all that all the time, but it's a matter of _how_ the applications are executing. If your machine is barely capable of servicing one of those applications, you either need to fix or upgrade your hardware and/or OS... or get something that better manages it.
Having a box peaking non-stop is always going to cause problems regardless of hyperthreading or not. I don't see how your argument holds up in the face of a pretty common case I outlined. Even if you exclude the compilation, that's still an amazing number of context switches to handle the network, writing, and reading between frame rendering.
If you have a tool such as vmstat, and you observe the actual context switches, they generally aren't in the thousands per second. If you're not doing excessive switching then HT gives you almost nothing. If you are doing thousands per second normally, you need to uninstall the neato-apps, uninstall the whiz-bang fancy features, and then test again.
Most apps are not starving for time. Most of the cases I've seen where they are stem from DMA issues on hard disks and burners, not because you can't service the applications fast enough.
vmstat -- Watch the CS
uptime -- watch the load
top -- watch the processor usage
Or use whatever windows has for these. The task manager can do the latter two... The CS, I'm not sure of.
If you have high CS, high load, and low CPU... There is a bounding issue that isn't processing -- It's most likely disk (or some other IO device) based.
If the CS is high, the load is low... Then maybe you have some weight to you comments. HT might help this case significantly... but I'd suggest you try another app that utilizes larger slices.
If the CPU is high, and the load is high, you have some contention issues that you need to work out regardless of what is happening with CS.
Got you covered, I ordered mine last night.
You can thank me later.
Techreport's test is from feburary 20th, and shows somewhat the same - AMD64 is king of gaming, and doing one thing at a time - Intel is good at encoding etc, and thanks to HT, doing more than one thing at a time - I would prefer AMD for now (I got a Prescott 550 3.4GHz - I wish I had an AMD). :D
Ain't it great with competition?!
Any technology distinguishable from magic, is insufficiently advanced.