0

I've been chugging through my homework and this question keeps haunting me. Am I doing it all wrong?

I've taken the stance that the 'block address' is the part of these 32 bit reference addresses (given in most problems) that do not include the least-significant bits used as the byte offset (the part useful for selecting a word (or byte) within a block.

To demonstrate, say we are using a direct-mapped cache and given a 32 bit reference address such as 0000 0000 0000 0000 1111 1100 0001 1100 and that the tag is 24 bits, the index is 6 bits, and the offset is 2 bits.

My textbook (Computer Organization and Design: MIPS Edition) states that the block a given address maps to in a direct mapped cache is found by (block address) modulo (number of blocks in the cache). If I include the offset in this calculation as part of the 'block address' the resulting block I would indicate a hit/miss/store from would be very different than if I hadn't.

Thus, it would put me at ease to know if anyone knew if 'block address' was just another term for 'address' (the entire 32 bit reference) or if it means what I think it means which is the concatenation of the Tag and the Index (nothing more). Anyone know?

Improve this question

1 Answer 1

Reset to default
2

Since mips uses words for addressing all addresses are essentially 30 bits long with 2 additional bits set to 0 (the byte offset at the end). The block-offset is the next 2 bits followed by the block address:

XXXX XXXX XXXX XXXX XXXX XXXX XXXX WW00

  • 00 this is the word-offset, mostly useless in mips as you usually don't use bytes
  • WW this is the block-offset, used to identify which word to choose (assuming the block is 4 words long)
  • XX.. the block address, this is the direct address of the memory block which contains 4 words and thus 4x4 = 16 bytes.

When you talk about the block address you talk about the first 28 bits of the address as the rest are only offsets for other applications.

  • byte address - all 32 bits
  • word address - the first 30 bits
  • block address - the first 28 bits (or less if you choose to have bigger blocks)

Tags and indexes only come into play when talking about a cache, at this point you look at the block address (in this case 28 bits) and choose your tag and index from this part only, the offsets are only used after cache access.

  1. Get tag and index from block address
  2. Take block-offset to get the word (done using a 32x4-1-MUX for example)
  3. Move word into destination register
Improve this answer
Sign up to request clarification or add additional context in comments.

1 Comment

mostly useless in mips as you usually don't use bytes. Strings in ASCII and UTF-8 are arrays of single bytes. Not exactly rare in typical real computer programs. Unaligned words are less commonly used, and until MIPS32r6 weren't guaranteed to be supported at all. (MIPS32r6 removed the special unaligned-load instructions (like lwl/r left/right half) and requires that regular lw / sw work on unaligned addresses. en.wikipedia.org/wiki/MIPS_architecture#MIPS32/MIPS64_Release_6). Other than that, +1 though.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.