Add short presentation about RISC-V hardware for Koji

diff --git a/2023-hardware/.gitignore b/2023-hardware/.gitignore
new file mode 100644 (file)
index 0000000..b25c15b
--- /dev/null
+++ b/2023-hardware/.gitignore
@@ -0,0 +1 @@
+*~

diff --git a/2023-hardware/koji-hardware.txt b/2023-hardware/koji-hardware.txt
new file mode 100644 (file)
index 0000000..2787e60
--- /dev/null
+++ b/2023-hardware/koji-hardware.txt
@@ -0,0 +1,125 @@
+Koji is Fedora's build system
+
+http://fedora.riscv.rocks/koji/
+https://koji.fedoraproject.org/koji/
+
+Doesn't use cross-compilation, and not possible with current design.
+We must find a way to run a full build system somehow.
+
+Discussion of primary architecture.
+
+
+
+
+## What hardware should we choose for RISC-V as a primary architecture?
+
+* Milk-V Pioneer (SG2042)
+* Sophgo 2U server (SG2042)
+
+  - Possible performance and quality issues [unverified]
+  - Expensive
+  - Likely to end up as e-waste soon.
+
+* SiFive Unmatched
+
+  - Slow
+  - Doesn't support any recent extensions, esp. V, H
+  - No BMC, but might use BMC-on-PCIe card
+
+* [Mystery manufacturer] future development board (under NDA)
+
+* Single board computers (SBCs)
+
+  - VisionFive2 (StarFive JH7xxx), Lichee Pi 4A, ...
+  - Slow
+  - Limited memory, cores
+  - Reliability issues
+  - Upfront & ongoing engineering headaches integrating into a 19" rack
+
+* QEMU
+
+  - Performance numbers below, but not great
+  - Really using x86-64 hardware
+  - Hardware (x86-64) is well known and easy to manage
+  - No "e-waste", servers can be repurposed
+  - Supports all the latest extensions
+  - Supports any amount of RAM, large numbers of vCPUs
+  - We can add new extensions and fix bugs relatively easily
+
+
+
+
+
+
+## QEMU performance numbers
+
+(Times in seconds, taken from recent Fedora Koji builds)
+
+               binutils     openssl      python3.12   mingw-gcc
+
+i686           1589         1577         3411         4292
+
+x86-64         1419         1172         2462         1827
+
+aarch64        1573          811 (?)     1845         2521
+
+ppc64le        2165         1291         3073         4388
+
+s390x          2553         1380         1984 (?)     6824
+
+               ------------ ------------ ------------ ------------
+
+  [qemu-system-riscv64 16 cores, 16 GB; on AMD Ryzen 9 7950X server]
+
+qemu riscv64
+  (LTO)        4493         3052        14502        12428
+                 +217%        +160%       +489%        +580%
+
+  (no LTO)     3267         1351         6353        (failed)
+                 +130%        +15%        +158%
+
+VisionFive 2
+  (LTO)        7202         8823
+                 +408%        +653%
+
+  (no LTO)     3274         2059
+                 +130%        +75%
+
+
+
+## QEMU observations
+
+"TCG" is the name for QEMU's software emulation, eg. RISC-V fully
+emulated guest on x86-64 host.
+
+Works using Translation Blocks (TBs) which translate basic blocks of
+guest code.
+
+Well understood (by me), easy to fix simpler issues.
+
+ - I posted a patch yesterday which gets ~ +6% performance gain
+
+A few tips to make TCG run (a bit) fast(er):
+
+ - Compile with -march=native (+4%)
+
+ - Profile with perf
+
+ - Don't overprovision host CPUs
+
+    * However pinning vCPUs to pCPUs didn't really help
+
+ - Give it plenty of guest & host RAM
+
+    * Measured memory overhead on host is up to 40% after running
+      for some time
+
+    * Host TBs track guest page cache; as long as a translated
+      executable remains in the guest page cache, it will not be
+      retranslated
+
+ - Don't restart the VM
+
+ - Software TLB
+
+ - Fast vs slow jumps