|
|
c2684b995b
|
Make time_tsc refer to the entire runtime again
Also correct tsc errors by adding the cyc duration
|
2025-09-12 12:44:41 +02:00 |
|
|
|
f9b0bf0b5f
|
Correct for errors in the current tsc by using cyc
|
2025-09-12 11:53:11 +02:00 |
|
|
|
2732383465
|
Adjust tsc calculation
Since we only want to record the time when benchmarking
|
2025-09-12 11:04:02 +02:00 |
|
|
|
718a7fe078
|
Add nyx benchmark without pt
|
2025-09-10 16:14:30 +02:00 |
|
|
|
c2e432c219
|
Benchmark nyx and baseline
|
2025-09-05 15:00:47 +02:00 |
|
|
|
d14e0fe3a0
|
Calculate the bus frequency at runtime
The 100mHz value was guaranteed by intel for this CPU,
but this approach is more reliable
|
2025-09-04 18:32:58 +02:00 |
|
|
|
f072c0ce53
|
Use correct conversion for time values
Two errors I made:
* cycle durations are not constant, their value relies on the cbr
* tsc frequency was not exactly 2.7Ghz, now calculated at runtime
|
2025-09-04 18:11:20 +02:00 |
|
|
|
10ca8f89f6
|
Track total time using tsc packets and total internal time with cycles.
Something is still wrong though, the tsc time is somehow less than the cyc time
|
2025-09-04 16:31:17 +02:00 |
|
|
|
2e5af76dbb
|
Track time in main and time disabled
|
2025-09-03 12:03:23 +02:00 |
|
|
|
6afa71cc35
|
Use ptwrite to capture runtime more accurately
|
2025-08-20 14:15:46 +02:00 |
|
|
|
e2d9fadcc8
|
Analyze trace dump for time information
|
2025-08-12 15:40:52 +02:00 |
|
