1: we manage to issue 33 operations / sec. This is easily a world record :) The way we do this is covered in the Instruction Encoding talk. We could conceivably push it further, but its diminishing returns. We can have lots of cores too.
2: its process agnostic; the dial goes all the way up to 11
3: the on-chip cache is much quicker than conventional architectures as the TLB is not on the critical path and we typically have ~25% fewer reads on general purpose code due to backless memory and implicit zero. The main memory is conventional memory, though; if your algorithm is zig zagging unpredictably through main memory we can't magic that away
I would really like to know your reasoning that the TLB is a major bottleneck in conventional CPUs. CPUs execute a TLB lookup in parallel with the cache, so there is usually no latency except on a TLB miss.
Basic research on in-memory databases suggest eliminating the TLB would improve performance only by about 10%, this certainly isn't a realistic use case and most of the benefits can be obtained simply by using larger pages. So I don't really know where your claims about 25% fewer reads is coming from in relation to simply getting rid of virtual memory.
But while I think the Mill people are overselling the direct performance benefits here, the single address space lets them do a lot of other things such as backing up all sorts of things to the stack automatically on a function call and handling any page fault that results in the same way that it would be handled if it was the result of a store instruction. And I think they're backless storage concept requires it too.
[1]http://en.wikipedia.org/wiki/CPU_cache#Address_translation [2] Unless you were to force the use of large page sizes, as some people suggest Apple might have done with their newest iPhone.
It was something about scopes mapping very well onto the Mill's "memory model."
I'm not quite up to that sort of analysis though, but I'm wondering if you see that too? If so, I would love to read more about that.
As far as I know, Elbrus 3M managed to achieve about 18 instructions per clock cycle, with VLIW and highly complex register file, whose design slowed overall clock frequency to about 300MHz on 0.9um process. To get everything in comparison, plain Leon2 managed to get about 450MHz in the same process, without any tweaks and hand work and Leon2 is not a speed champion.
So the questions is: do you have your world record in simulation or in real hardware like FPGA?
33 ops / cycle, sustained. Last night we also published an example list of the FU mix on those pipelines here: http://ootbcomp.com/topic/introduction-to-the-mill-cpu-progr...
Depends on how you define simplicity, really. Writing a good back-end for this architecture is likely to be very challenging.
What challenges would you think a back-end developer would face?
There's more information about memory in the talk at http://ootbcomp.com/topic/memory/
Would be interesting to find out:
1. How high that degree of parallelism can be pushed, are we talking about tens or hundreds of pipelines?
2. What frequency this will operate at?
3. What is up with RAM? I saw nothing about memory, with lots of pipelines it is bound to be memory bound.