tag:blogger.com,1999:blog-3155908228127841862.post6347378488892830225..comments2023-06-28T10:04:44.463-06:00Comments on The Perils of Parallel: Oh, for the Good Old Days to ComeGreg Pfisterhttp://www.blogger.com/profile/12651996181651540140noreply@blogger.comBlogger4125tag:blogger.com,1999:blog-3155908228127841862.post-74503559339620940882009-12-05T21:32:44.789-07:002009-12-05T21:32:44.789-07:00For me, the Ho. Ly. Crap apps on multicore are dma...For me, the Ho. Ly. Crap apps on multicore are dmake and 7zip - in fact, I use 7zip (on Windows it integrates cleanly with Explorer, CLI on other OSs) whenever possible now. Compressing and decompressing are fast as blazes, as is compiling. As for zippier, if you use Eclipse you'll definitely notice a difference.<br /><br />Photoshop CS4 really flies. I can't speak for Adobe's other apps, but the integration with Bridge is threaded, so batch-processing images uses all cores. Nvidia's drivers are tailored to the 3D capabilities of the CS4 suite (on all cards, not just Quadro), and I'm running a GTX-285 GPU. I don't do 3D professionally, but playing around with it is surreal compared to my previous desktop-PC experience.<br /><br />I realize this isn't most peoples' experience with a multi-core upgrade. But I do believe that it will be, within a few short years, as more applications catch up.<br /><br />It's been a few months now and I'm still grinning like an idiot, since it turns out most of the major applications I use were already good-to-go for multicore. The five years since my last PC upgrade feel like a decade, performance-wise.Eric J. Bowmanhttp://example.org/noreply@blogger.comtag:blogger.com,1999:blog-3155908228127841862.post-38922472674397409552009-12-05T13:56:32.318-07:002009-12-05T13:56:32.318-07:00Well, try the Intel G2 SSD on your system and yo...Well, try the Intel G2 SSD on your system and you will find that everything is _much_ faster. In fact i have never experienced an upgrade that made such a big difference as this SSD has.Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-3155908228127841862.post-79070027198488896482009-12-04T14:52:35.696-07:002009-12-04T14:52:35.696-07:00Hi, Noah.
Yes, certainly disk seek time has kept ...Hi, Noah.<br /><br />Yes, certainly disk seek time has kept PCs (& Macs) from being zippier, and I hope SSDs do help. It alone might make a bigger real difference than several old generations of silicon.<br /><br />But there was still zip despite disk relative slowness in the 90s and early 2000s. That zip was lost. We need post-Moore programs to get it back.<br /><br />GregGreg Pfisterhttps://www.blogger.com/profile/12651996181651540140noreply@blogger.comtag:blogger.com,1999:blog-3155908228127841862.post-19942581419142343772009-12-04T10:25:00.139-07:002009-12-04T10:25:00.139-07:00Greg wrote:
---
Of course, the culprit denying us...Greg wrote:<br /><br />---<br />Of course, the culprit denying us<br />this small pleasure has been the<br />flattening of single-thread<br />performance wrought by the half<br />death of Moore's Law. <br />---<br /><br />Hi Greg! I'm thinking that a significant factor might also be the long-flat curve of disk seek time vs. calendar year. Certainly the Windows systems I run seem to be disk-bound when they're at their slowest (not to say that flattening of the single core curve isn't an important and less well understood issue too.)<br /><br />Turning around, one might say it's interesting that the ratio of disk performance to CPU performance isn't changing as quickly as it once was (I.e. because neither of them is progressing very much! Maybe SSDs will invert the relative rates of progress; not sure).<br /><br />Cheers.<br /><br />NoahNoah Mendelsohnhttp://www.arcanedomain.comnoreply@blogger.com