Silicon Investor (SI) -- The First Internet Community

STOCKTALK

We've detected that you're using an ad content blocking browser plug-in or feature. Ads provide a critical source of revenue to the continued operation of Silicon Investor. We ask that you disable ad blocking while on Silicon Investor in the best interests of our community. If you are not using an ad blocker but are still receiving this message, make sure your browser's tracking protection is set to the 'standard' level.

Politics : Formerly About Advanced Micro Devices -- Ignore unavailable to you. Want to Upgrade?

To: Process Boy who wrote (55753)	4/16/1999 6:40:00 PM
From: Scumbria	Respond to of 1579026

PB, How reliable of a gauge would these simulated values be? do they take into account redundancy features of the respective caches? Just curious. Cache hit/miss rates can be reliably simulated from a trace simulator. Unfortunately that is the only reliable feature. Scumbria

To: Process Boy who wrote (55753)	4/16/1999 6:48:00 PM
From: Tenchusatsu	Read Replies (1) \| Respond to of 1579026

<How reliable of a gauge would these simulated values be? do they take into account redundancy features of the respective caches? Just curious.> I don't think these simulated values are reliable. There are way too many variables, like cache-replacement policies, the number of ways per set (2-way, 4-way, direct-mapped), the traffic pattern generated by the decoders and execution units, and so on. But in general, they do reflect the expectation that the cache miss ratios for Coppermine should be higher than the miss ratios for K7. This is so because the K7 has larger caches than Coppermine. This has little to do with how fast those caches are, however. I'm also very surprised that there were cases where the IPC (instructions per cycles) of Coppermine are higher than that of K7, given that the K7 has more execution units and deeper buffers than Coppermine. That's why I think those IPC values are also suspect. Tenchusatsu