| 1 |
On Jul 28, 2012 8:03 AM, "Michael Mol" <mikemol@×××××.com> wrote: |
| 2 |
> |
| 3 |
> On Fri, Jul 27, 2012 at 8:30 PM, microcai <microcai@×××××××××××××.org> |
| 4 |
wrote: |
| 5 |
|
| 6 |
--- >8 Major Snippage |
| 7 |
|
| 8 |
> > As far as I can tell, AMD chip suffered with a lot of I/O. Their |
| 9 |
> > Hyper-transport seems not competitive with Intel's ring bus |
| 10 |
> |
| 11 |
|
| 12 |
Wasn't Intel's answer to HyperTransport is the QuickPath bus? IIRC, the |
| 13 |
ring bus is internal to a processor. (I could be wrong, though). |
| 14 |
|
| 15 |
> (please don't top-post, especially if the thread's already been |
| 16 |
> primarily organized as bottom-post) |
| 17 |
> |
| 18 |
> I hadn't read that, but remember that HyperTransport is intended for a |
| 19 |
> mesh architecture. In single-CPU systems, you'll only have one HT |
| 20 |
> link, the link between your CPU and your north bridge. In multi-CPU |
| 21 |
> systems, you'll have additional links between the CPUs. In systems |
| 22 |
> with many CPUs, you may even have a fully-connected mesh. |
| 23 |
> |
| 24 |
> The I/O characteristics will greatly depend on the topology of your |
| 25 |
network. |
| 26 |
> |
| 27 |
> That said, HyperTransport may just be getting old; when it came out, |
| 28 |
> it (and AMD's crossbar switch for memory management) beat the pants |
| 29 |
> off of Intel's SMP solution. Intel's solution ran at lower and lower |
| 30 |
> clock rates the more CPUs you added, and their first pass at multicore |
| 31 |
> gave each core its own port onto the memory bus, with predictably poor |
| 32 |
> results. Intel's had plenty of time to catch up, but with their |
| 33 |
> price-per-part, it's taken me a long time to pay much attention. |
| 34 |
> |
| 35 |
|
| 36 |
Again, I might be mistaken, but IIRC HyperTransport's throughput depends on |
| 37 |
how many channels are provided, so there's no theoretical limitation to its |
| 38 |
throughput, just practical considerations. (E.g., tracing issues). |
| 39 |
|
| 40 |
> (It also doesn't help that Jon "Hannibal" Stokes stopped writing |
| 41 |
> detailed technical articles for Ars Technica; I sincerely miss him and |
| 42 |
> the precision and clarity of his writing on such arcane subjects.) |
| 43 |
> |
| 44 |
|
| 45 |
That makes the two of us bro... BTW, my handle there's "pepoluan", just in |
| 46 |
case you see it in the forums. |
| 47 |
|
| 48 |
Rgds, |
Archives