Messages in this thread | | | Date | Wed, 18 Mar 1998 10:03:56 +0000 (GMT/BST) | From | Mike Jagdis <> | Subject | Re: BogoMIPS revisitied...MMX |
| |
On Tue, 17 Mar 1998, George Bonser wrote:
> Here is an interesting array: > > AMD K5-100: 199.07 BMPS > INTEL P-60: 23.96 BMPS > Cryix 6x86 P150+ (at 120MHz): 52.33 BMPS > INTEL 486-DX266: 33.08 BMPS. > > Why is the AMD 100MHz CPU 4x faster (in BogoMIPS) than a Cyrix running a > 20% higher clock speed? I would say that the AMD CPU Microcode is > optimized more toward the type of instruction that the bogomips loop > executes. Why is it nearly 10x faster than the Intel chip?
I guess the AMD is better optimised for an empty loop. I have the feeling that the 6x86 only does branches in the X pipeline. If the AMD does them in both pipelines the difference in an empty loop could be a substantial part of the total. The 486 simply doesn't have the pipelining, register renaming, branch prediction etc. so I wouldn't expect high numbers. The P60 is similarly, but not as comprehensively, held back.
> I dunno, but I think I will buy more AMD chips ;)
Huh? You want to run empty loops? If you don't know why you see a difference in a very specialized piece of code you shouldn't assume the difference scales to general code. *Everyone* optimises for the common benchmarking code. But this is only good if you want to run benchmarks fast :-).
Mike
-- .----------------------------------------------------------------------. | Mike Jagdis | Internet: mailto:mike@roan.co.uk | | Roan Technology Ltd. | | | 54A Peach Street, Wokingham | Telephone: +44 118 989 0403 | | RG40 1XG, ENGLAND | Fax: +44 118 989 1195 | `----------------------------------------------------------------------'
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu
| |