Andrew Waterman
6510f020c7
fix deadlock in coherence hub
2012-08-03 19:00:03 -07:00
Andrew Waterman
e3726c4db0
fix control bug in LLC
...
structural hazard on tag ram caused deadlock
2012-08-03 18:59:37 -07:00
Andrew Waterman
def913096e
pipeline LLC further
2012-07-31 17:45:14 -07:00
Andrew Waterman
3a8f3e0de5
further pipeline the LLC
2012-07-30 20:12:11 -07:00
Andrew Waterman
be4fa936dd
fix PriorityEncoderOH bug
2012-07-30 18:28:54 -07:00
Andrew Waterman
130fa95ed6
expand HTIF's PCR register space
2012-07-27 14:52:39 -07:00
Andrew Waterman
9c50621a19
remove chip-specific uncore gunk
2012-07-26 03:26:52 -07:00
Andrew Waterman
a5bea4364f
memory system bug fixes
2012-07-26 00:05:21 -07:00
Andrew Waterman
177dbdadd9
merge HTIF port and backup memory port
2012-07-25 00:18:02 -07:00
Andrew Waterman
a21c355114
fix htif split request/response
2012-07-23 17:15:16 -07:00
Andrew Waterman
938effc053
don't dequeue probe queue during reset
2012-07-22 21:05:52 -07:00
Andrew Waterman
f42c6afed2
decouple all interfaces between tile and top
...
also, add an "incoherent" bit to tilelink to indicate no probes needed
2012-07-17 22:55:40 -07:00
Andrew Waterman
4e44ed7400
allow back pressure on IPI requests
2012-07-17 22:55:40 -07:00
Andrew Waterman
5035374f36
update to new chisel
2012-07-08 17:59:41 -07:00
Andrew Waterman
4e5f874266
update to new chisel/hwacha
2012-06-08 00:13:14 -07:00
Andrew Waterman
943b6d0616
remove debug println
2012-06-06 02:48:48 -07:00
Andrew Waterman
7f6319047e
update to new scala/chisel/Mem
2012-06-06 02:47:22 -07:00
Andrew Waterman
faee45bf4c
fix setpcr/clearpcr not writing rd
2012-05-21 07:25:35 -07:00
Andrew Waterman
a2f6d01c1b
add programmable coreid register
2012-05-09 03:09:22 -07:00
Andrew Waterman
e0e1cd5d32
add IPIs and an IPI test
...
IPIs are routed through the HTIF, which seems weird, but that makes it
so cores can bring each other out of reset with IPIs.
2012-05-08 22:58:00 -07:00
Andrew Waterman
b851f1b34c
support maximum-MTU HTIF packets
2012-05-03 21:11:43 -07:00
Andrew Waterman
171c87002e
reduce HTIF clock divider for now
2012-05-03 04:21:11 -07:00
Andrew Waterman
e1f9dc2c1f
generalize page table walker
...
also, don't instantiate vitlb when !HAVE_VEC
2012-05-03 02:29:09 -07:00
Andrew Waterman
2d4e5d3813
fix pseudo-LRU verilog generation bug
2012-05-02 19:31:31 -07:00
Andrew Waterman
65ff397122
improved instruction decoding
...
it now makes use of don't-cares by performing logic minimization
2012-05-01 20:16:36 -07:00
Andrew Waterman
4cfa6cd9a8
force Top.main's return type to Unit
2012-05-01 19:55:16 -07:00
Andrew Waterman
5819beed64
use parameterized FP units
2012-05-01 01:25:43 -07:00
Andrew Waterman
eafdffe125
simplify page table walker; speed up emulator
2012-05-01 01:24:36 -07:00
Andrew Waterman
c13d3e6f88
fix probe tag read-modify-write atomicity violation
2012-04-26 02:29:31 -07:00
Andrew Waterman
66f86a2194
use pseudo-LRU replacement for TLBs
2012-04-26 02:29:30 -07:00
Andrew Waterman
a0378c5d2f
remove faulting TLB entry after page fault
...
this vastly reduces the frequency with which the TLB must be flushed
2012-04-26 02:29:30 -07:00
Andrew Waterman
6d8fc74378
fix DTLB permissions bug
2012-04-26 02:29:30 -07:00
Andrew Waterman
fb4408b150
fix AMO replay/coherence deadlock
2012-04-15 22:56:02 -07:00
Andrew Waterman
724735f13f
fix writeback bug
2012-04-13 03:16:48 -07:00
Andrew Waterman
00d934cfac
fix coherence bugs in cache
2012-04-12 21:57:37 -07:00
Andrew Waterman
c0ec3794bf
coherence mostly works now
2012-04-10 02:22:45 -07:00
Andrew Waterman
aee9378712
fix coherence bug with multiple probe replies
2012-04-09 21:40:35 -07:00
Andrew Waterman
7f254d9670
refine FP bugfixes
2012-04-01 14:52:33 -07:00
Andrew Waterman
a09e8d1c55
remove I$ prefetcher for now
...
there's a bug in it, and I don't have time to fix it at the moment.
2012-03-27 15:43:56 -07:00
Andrew Waterman
452876af37
fence on vvcfg; implement fence.v.g correctly
2012-03-27 14:49:00 -07:00
Andrew Waterman
6bda8674bd
no dessert tonight :(
2012-03-26 23:50:09 -07:00
Andrew Waterman
e2fe525fb6
remove bug from dessert
2012-03-26 14:18:57 -07:00
Andrew Waterman
5f53cd4ac1
reduce HTIF width
2012-03-25 23:49:59 -07:00
Andrew Waterman
ef505de017
reduce HTIF width
2012-03-25 23:49:45 -07:00
Andrew Waterman
31f0b600fd
add dessert
2012-03-25 23:03:20 -07:00
Andrew Waterman
1666d3fbd7
loop host.in to host.out during reset
2012-03-25 21:45:10 -07:00
Andrew Waterman
f62a02ab54
remove dumb stuff in top.scala
2012-03-25 21:30:01 -07:00
Andrew Waterman
88bf8a4f23
add mem serdes unit
2012-03-25 17:03:58 -07:00
Andrew Waterman
7fa93da4f5
add backup memory port (disabled for now)
2012-03-25 15:49:32 -07:00
Andrew Waterman
86d56ff67b
refactor cpu/i$/d$ into Tile (rather than Top)
2012-03-24 16:57:28 -07:00
Andrew Waterman
3a487ac89b
improve htif<->pcr interface
2012-03-24 16:57:28 -07:00
Andrew Waterman
54fa6f660d
new supervisor mode
2012-03-24 13:03:31 -07:00
Andrew Waterman
bd27d0fab2
can now take interrupts on stalled instructions
2012-03-19 01:02:06 -07:00
Andrew Waterman
2ed0be65f9
fix RRArbiter
2012-03-19 00:19:33 -07:00
Andrew Waterman
c4a91303fb
update vector fence names and encoding
2012-03-18 20:42:38 -07:00
Andrew Waterman
a47eeb9571
retime D$ bypass into beginning of EX stage
2012-03-16 18:35:54 -07:00
Andrew Waterman
6c26921766
reduce D$ critical path through page table walker
...
costs an extra cycle per page table level to resolve a TLB miss. too bad.
2012-03-16 18:35:54 -07:00
Andrew Waterman
f0157b9e2a
fix coherence bug
...
popping wrong store dependence queue
2012-03-16 01:24:07 -07:00
Andrew Waterman
cfca2d1411
clean up cache interfaces; avoid reserved keywords
2012-03-16 00:44:16 -07:00
Andrew Waterman
820884c7e6
fix probes for smaller cache sizes
...
address bits (pgidx_bits-1,taglsb) were omitted from tag checks.
2012-03-15 23:08:30 -07:00
Andrew Waterman
4684171ac6
fix fence.i for associative caches
2012-03-15 21:23:21 -07:00
Andrew Waterman
2b0bc8df2b
use divided clk for htif. UDPATE YOUR FESVR
...
by default, we now load programs via a backdoor, because otherwise
it takes too long to simulate.
2012-03-15 18:36:51 -07:00
Andrew Waterman
7dde7099d2
use broadcast hub and coherent HTIF
2012-03-14 16:44:35 -07:00
Andrew Waterman
ab6c9350db
fix minor coherence bugs
2012-03-13 19:10:54 -07:00
Andrew Waterman
1788c34113
parameterize broadcast hub by # of tiles
2012-03-13 17:12:01 -07:00
Andrew Waterman
1492457df5
add probe replies to HTIF
2012-03-13 16:56:47 -07:00
Andrew Waterman
b0f798962c
add probe unit
2012-03-13 16:43:51 -07:00
Andrew Waterman
d76b05bde1
fix way selection on D$ write upgrades
2012-03-13 02:21:02 -07:00
Andrew Waterman
fd29e00db0
support non-power-of-2 queue sizes
...
need to manually wrap queue pointers.
2012-03-13 01:58:28 -07:00
Andrew Waterman
8ffdac9526
fix D$ store-upgrade bug
...
loads to the same address as stores that cause an upgrade
could return the old value
2012-03-10 15:50:10 -08:00
Andrew Waterman
4f4b990a4f
fix null hub store ack bug
2012-03-10 15:19:12 -08:00
Andrew Waterman
7eb73c325e
fix signedness of zero fmul results
...
We were using the FMA unit to compute rs1 * rs2 + 0.0 for fmul,
which incorrectly computes +0.0 when rs1 * rs2 == -0.0. Now we
add -0.0 if rs1*rs2 is negative.
2012-03-10 00:21:51 -08:00
Andrew Waterman
e3a68848e0
fix D$ critical paths and fix verilog build
2012-03-09 20:02:51 -08:00
Andrew Waterman
85504f0ddc
fix bug in fence.i and improve test
2012-03-09 03:26:05 -08:00
Andrew Waterman
766bac88f8
refactor D$ writebacks and flushes
...
MSHRs now arbitrate for writebacks and handle flushes.
2012-03-09 02:55:46 -08:00
Andrew Waterman
ff2e47f380
Merge branch 'master' of github.com:ucb-bar/riscv-rocket
2012-03-09 02:08:55 -08:00
Andrew Waterman
5a7c5772a8
clearly distinguish PPN and cache tag
2012-03-07 23:11:17 -08:00
Andrew Waterman
941873bad1
coherence hub fixes
2012-03-07 21:03:44 -08:00
Andrew Waterman
c09eeb7fd2
fix D$ next-state logic
...
it was using the CPU command from the wrong pipeline stage,
which was a don't-care with ThreeStateIncoherence.
2012-03-07 01:42:08 -08:00
Andrew Waterman
a0c9452b86
change D$ to use FourStateCoherence protocol
...
instead of ThreeStateIncoherence.
2012-03-07 01:26:35 -08:00
Andrew Waterman
6e2610b0ad
fix Mux1H for bundles
2012-03-06 23:38:36 -08:00
Andrew Waterman
499c5b4a2e
automatically infer MEM_TAG_BITS
2012-03-06 15:49:28 -08:00
Andrew Waterman
6e16b04ada
implement transaction finish messages
2012-03-06 15:48:08 -08:00
Andrew Waterman
5f33ab24b0
fix merge conflict
...
oops :(
2012-03-06 02:02:53 -08:00
Andrew Waterman
5f12990dfb
support memory transaction aborts
2012-03-06 00:35:02 -08:00
Andrew Waterman
1e1926ce63
flip direction of ioPipe to match ioDecoupled
2012-03-02 16:18:32 -08:00
Andrew Waterman
6d03d75835
improve D$ internal interfaces
2012-03-01 20:20:15 -08:00
Andrew Waterman
28cacd953f
D$ cleanup - merge ReplayUnit and MSHRFile
2012-03-01 19:30:56 -08:00
Andrew Waterman
52101373e0
clean up D$ store data unit
2012-03-01 19:20:00 -08:00
Andrew Waterman
c38065d0e8
clean up priority encoders
2012-02-29 16:13:14 -08:00
Andrew Waterman
b9ec69f8f5
add new Queue singleton
2012-02-29 14:21:42 -08:00
Andrew Waterman
012da6002e
replace tile memory interface with ioTileLink
...
work in progress towards coherent HTIF. for now, requests
are incoherently passed through a null coherence hub.
2012-02-29 03:10:47 -08:00
Andrew Waterman
c99f6bbeb7
separate memory request command and data
...
also, merge some VLSI/C++ test harness functionality
2012-02-28 19:06:23 -08:00
Andrew Waterman
2b1c07c723
replace ioDCache with ioMem
2012-02-27 18:36:09 -08:00
Andrew Waterman
1d41a41afa
remove extraneous constants
2012-02-27 17:49:48 -08:00
Andrew Waterman
6e706c7c74
fix yet another AMO-related replay bug
2012-02-26 20:20:45 -08:00
Andrew Waterman
e12b9eae93
remove ext_mem interface
...
hindsight is 20/20
2012-02-26 18:53:39 -08:00
Andrew Waterman
2d04664a98
simplify cpu-cache interface
2012-02-26 18:26:29 -08:00
Andrew Waterman
ad713a5d83
fix icache ram depth; new chisel
2012-02-26 17:51:46 -08:00
Andrew Waterman
4121fb178c
clean up mul/div interface; use VU mul if HAVE_VEC
2012-02-24 19:22:35 -08:00