|バージョン||0.7.4.9478 0.7.4.9477 0.7.3.9475 0.7.3.9474 0.7.3.9472 0.7.3.9471 0.7.2.9469 0.7.2.9468b 0.7.2.9468 0.7.1.9466b 0.7.1.9466|
2017/12/03 ver 0.7.4.9478
December 3, 2017)
Windows: Version 0.7.4 Build 9478 (Released: December 2, 2017)
Linux : Version 0.7.4 Build 9478 (Released: December 2, 2017)
AV False Positives: (October 24, 2017)
It has come to my attention that the "y-cruncher.exe" launcher binary in Windows has been getting flagged by some virus scanners. This has been the case for at least the last several releases.
This is currently under investigation. Unfortunately, I'm not an expert in this field. So I don't really know what exactly is tripping up the AV heuristics. At this time, the only fix I've found will break compatibility with older versions of Windows.
In case anybody is willing to help, the source code for the launcher binary has been released on GitHub. Compiling it using the provided Visual Studio project will produce the binary that turns up false positive under said virus scanners.
Latest Releases: (December 2, 2017)
2017/10/15 ver 0.7.4.9477
October 14, 2017)
Windows: Version 0.7.4 Build 9477 (Released: October 14, 2017)
Linux : Version 0.7.4 Build 9477 (Released: October 14, 2017)
Version 0.7.4: (October 14, 2017)
A new version is out with some changes aimed at addressing the memory bottleneck on Ryzen 7 and Skylake X.
The stress-tester has been revamped. There are two new tests designed to stay within cache to avoid bottlenecking on memory access.
Some minor optimizations have been made to reduce the pressure on memory bandwidth.
A bug that caused excessive memory usage with large task decompositions has been fixed. It is now possible to fit the 25 billion digit Pi benchmark into 128 GB of memory on the high core-count processors with more than 16 logical cores.
Latest Releases: (October 14, 2017)
HWBOT Submitter v0.9.8.118.jar
Comparison Chart: (Last updated: October 14, 2017)
32 GB - 1600 MHz
2017/09/15 ver 0.7.3.9475
September 14, 2017)
Windows: Version 0.7.3 Build 9475 (Released: September 14, 2017)
Linux : Version 0.7.3 Build 9475 (Released: September 14, 2017)
I look forward to see what phantom throttling looks like under the VTune profiler.
August 24, 2017
August 23, 2017
Compute: 34.4 days
Verify: 141 days
4 x Xeon E5-4660 v3 @ 2.1 GHz - 1 TB
2 x Xeon X5690 @ 3.47 GHz - 128 GB
Latest Releases: (September 14, 2017)
2017/08/16 ver 0.7.3.9474
y-cruncher - A Multi-Threaded Pi-Program
From a high-school project that went a little too far...
By Alexander J. Yee
August 16, 2017)
Records Set by y-cruncher
2017/07/13 ver 0.7.3.9472
July 12, 2017)
Windows: Version 0.7.3 Build 9472 (Released: July 12, 2017)
Linux : Version 0.7.3 Build 9472 (Released: July 12, 2017)
EDIT (July 9, 2017):
I got asked about this, so here's some data showing the phantom throttling at stock settings. The pink entries are the ones with phantom throttling.
10 billion Hex-Digit of Pi - Plouffe's 4-term BBP Formula (y-cruncher v0.7.3)
Core i9 7900X - Gigabyte AORUS Gaming 7 (BIOS F7a)
All Stock Settings
2017/07/07 ver 0.7.3.9471
July 7, 2017)
Windows: Version 0.7.3 Build 9468 (Released: July 6, 2017)
Linux : Version 0.7.3 Build 9468 (Released: July 6, 2017)
Skylake X and AVX512: (July 6, 2017)
Not all Skylake X and Skylake Purley processors will have the full AVX512 capability.
"Phantom throttling" of performance when certain thermal limits are exceeded.
Memory bandwidth is a significant bottleneck.
Amdahl's law and other unknown scalability issues.
Not all Skylake X and Skylake Purley processors will have the full AVX512 capability:
While this reason doesn't apply to my system, it's worth mentioning it anyway.
However, that dedicated 512-bit FMA is only enabled on the Core i9 parts. The 6-core and 8-core Core i7 parts are supposed to have it disabled. Therefore they only have half the AVX512 performance.
It's worth mentioning that there is a benchmark on an engineering-sample 6-core Core i7 that shows full-throughput AVX512 anyway. However, engineering sample processors are not always representative of the retail parts.
"Phantom throttling" of performance when certain thermal limits are exceeded:
2017/06/04 ver 0.7.2.9469
June 3, 2017)
Windows: Version 0.7.2 Build 9469 (Released: June 3, 2017)
Linux : Version 0.7.2 Build 9469 (Released: June 3, 2017)
Core i7 5960X @ 4.0 GHz - 128GB DDR4 @ 2666 MHz - 16 HDs
Latest Releases: (June 3, 2017)
The parallel scalability problem in Linux has improved significantly with Ubuntu 17.04. But it's still not as good as in Windows. More testing is required.
1 Billion digits of Pi (times in seconds)
Intel Core i7 4770k
2017/03/21 ver 0.7.2.9468b
March 20, 2017)
Latest Releases: (March 20, 2017)
HWBOT Submitter v0.9.6.114.jar
Shigeru Kondo's Pi pages (dead)
Stu's Pi page (dead)
2017/03/14 ver 0.7.2.9468
March 13, 2017)
22.4 trillion digits - November 2016 (Peter Trueb)
Windows: Version 0.7.2 Build 9468 (Released: March 14, 2017)
Linux : Version 0.7.2 Build 9468 (Released: March 14, 2017)
Version 0.7.2 and AMD Zen: (March 14, 2017)
I went through a lot of trouble to do this in time for Pi day, but here it is. y-cruncher v0.7.2 has a new binary specifically optimized for AMD's Ryzen 7 processors.
As for the Zen architecture itself. Here are my (early) observations:
256-bit AVX instructions are handled efficiently enough that it seems to be beneficial to use 256-bit instructions when there's no overhead to doing so.
Memory is huge bottleneck. Latencies are very high and dual-channel memory is simply not enough to feed this much computational throughput.
For software developers, compiling code on the 1800X is about as fast as the 5960X at stock clocks. But the 5960X has much more overclocking headroom, so it ends up winning by around 15%. For a $500 processor, the R7 1800X is very impressive.
November 15, 2016
November 11, 2016
2016/10/31 ver 0.7.1.9466b
October 30, 2016)
Knights Landing Xeon Phi with AVX512: (October 10, 2016)
Core i7 5960X
2 x Xeon E5-2696 v4
Xeon Phi 7250
4.0 GHz (OC)
2016/09/17 ver 0.7.1.9466
September 16, 2016)
Windows: Version 0.7.1 Build 9466 (Released: September 16, 2016)
Linux : Version 0.7.1 Build 9466 (Released: September 16, 2016)
September 3, 2016
August 29, 2016
Compute: 48.6 days
Verify: 48.7 days
2 x Xeon X5690 @ 3.47 GHz
August 14, 2016
June 26, 2016