5 ms·
It probably does have an insane cache hierarchy, but in any case think of it this way: latency is going to be dramatically better to main memory on this machine
by rwallace 13y ago
It probably does have an insane cache hierarchy, but in any case think of it this way: latency is going to be dramatically better to main memory on this machine than to disk, which is what the alternative would be if you have a highly nonlocal workload.
- habosa 13y agoTrue, but gotta wonder how much better this latency is than a more reasonable amount of RAM backed by a super-fast SSD.
- rwallace 13y agoIt depends on locality I suppose. If your workload is spatially local, you're good on this machine and might also be good on a bunch of smaller boxes. If spatially nonlocal but temporally local, your solution might work well. If also temporally nonlocal, this machine might be your best bet; even if latency hurts bad, it might still be better than on any other hardware.
- frozenport 13y agoThere isn't a bus with the characteristics you mentioned. For example, SATA 3 saturates at 1 GBps. A few years ago I measured something like 50GBps on these guys. The real trick is to walk a graph with multiple processors so that you saturate the bus. That being said I liked the Yarc data architecture more.
- andruby 13y agoThat's why the fastest SSD's use PCI Express instead of SATA. [1] http://www.fusionio.com/products/iodrive-octal/ http://www.fusionio.com/products/iodrive-octal/ (6GB/s)
- frozenport 13y agoThere is an interesting story here, take a look at the benchmarks http://regmedia.co.uk/2011/04/07/ssd_write_drop_off.jpg http://regmedia.co.uk/2011/04/07/ssd_write_drop_off.jpg I did a bunch of work with SSDs and trying to get high throughput, at the end of the day I could touch ram at 10 megabytes per millisecond but only do IO at 2 megabyte per millisecond.