Daniel
0 Followers · 0 Following
  • Joined on 2023-09-27
Repositories
4
 Projects Packages Public Activity Starred Repositories
Daniel pushed to main at Daniel/master-thesis 2025-04-17 13:31:27 +02:00
c3f19d807a relwork: implemented Kronberger feedback
Daniel pushed to main at Daniel/master-thesis 2025-04-16 20:54:34 +02:00
6880c1ceb5 benchmarking: added uni performance test
Daniel pushed to main at Daniel/master-thesis 2025-04-15 19:32:54 +02:00
c62aff806a small updates and notes for further writing
Daniel pushed to main at Daniel/master-thesis 2025-04-13 14:20:25 +02:00
ef721b13e0 evaluation: updated notes for chapter
Daniel pushed to main at Daniel/master-thesis 2025-04-13 13:26:47 +02:00
a5c34a53b7 benchmarking: reverted previous; made interpreter use fast math
Daniel pushed to main at Daniel/master-thesis 2025-04-13 11:46:03 +02:00
6d6874c7ba benchmarking: added results for test 4 on uni-pc
Daniel pushed to 4-interpreter-using-int32 at Daniel/master-thesis 2025-04-13 11:43:32 +02:00
278a493595 benchmarking: tested int32 also on uni pc
Daniel pushed to 4-interpreter-using-int32 at Daniel/master-thesis 2025-04-13 11:34:29 +02:00
Daniel created branch 4-interpreter-using-int32 in Daniel/master-thesis 2025-04-13 11:34:29 +02:00
Daniel pushed to main at Daniel/master-thesis 2025-04-13 11:33:50 +02:00
af3b72f196 benchmarking: used int32 wherever possible; resulted in noticeable performance drop
Daniel pushed to main at Daniel/master-thesis 2025-04-12 16:22:23 +02:00
4c60331288 evaluation: added introduction text and made plan for additional text
Daniel created branch 3-tuned-blocksize-i128-t96 in Daniel/master-thesis 2025-04-12 13:22:20 +02:00
Daniel pushed to 3-tuned-blocksize-i128-t96 at Daniel/master-thesis 2025-04-12 13:22:20 +02:00
Daniel pushed to main at Daniel/master-thesis 2025-04-12 13:21:27 +02:00
101ccef67b benchmarking: tuned blocksizes; slightly improved performance; mostly improved standard deviation
Daniel commented on issue Daniel/master-thesis#14 2025-04-12 13:02:31 +02:00
Interpreter uses FP64

There are also registeres spilled into local memory. This should also be avoided

Daniel opened issue Daniel/master-thesis#14 2025-04-12 12:37:29 +02:00
Interpreter uses FP64
Daniel commented on issue Daniel/master-thesis#3 2025-04-12 12:36:44 +02:00
One thread per expression

Was not the case. Probably didn't understand it right

Daniel closed issue Daniel/master-thesis#3 2025-04-12 12:36:44 +02:00
One thread per expression
Daniel pushed to main at Daniel/master-thesis 2025-04-12 12:14:06 +02:00
c6e2ce47aa benchmarking: redid performance tests on uni pc
Daniel pushed to 1-performance-256-blocksize at Daniel/master-thesis 2025-04-12 10:36:01 +02:00