Begin forwarded message: From: "Vitali A. Morozov" <[email protected]<mailto:[email protected]>> Subject: Re: B0 KNL System Date: March 30, 2016 at 11:14:19 AM CDT To: "Fromkin, Russ" <[email protected]<mailto:[email protected]>>, Scott Parker <[email protected]<mailto:[email protected]>> Cc: "Allen, Benjamin S." <[email protected]<mailto:[email protected]>>, "Kumaran, Kalyan" <[email protected]<mailto:[email protected]>>, "Kirkendall, Keith G" <[email protected]<mailto:[email protected]>>, "Allison, CM" <[email protected]<mailto:[email protected]>>, "Arney, Rebecca" <[email protected]<mailto:[email protected]>>, Kalyanasundaram Kumaran <[email protected]<mailto:[email protected]>> Hi Russ, I am intensively benchmarking it for the last couple of days - it is too early to share much of the data that I am getting, I am afraid to make a mistake. However, of what I can share so far. I am very happy to see what is coming out after the BIOS update. The STREAM benchmark was fixed and I am measuring what I am expecting (not finished yet). HACCmk is running at 1.6 IPC (60% of peak)! I am very much surprised to see such a terrific match of the B0 hardware versus Sniper model. For HACCmk, Sniper was projecting n = 1488: 486725 cycles, IPC 1.5933, 49.9847% of peak For the same problem size and slightly modified coding, I have measured n = 1488: 501949 cycles, IPC = 1.5601, 57.5565% of peak. Amazing! Vitali On 03/30/2016 11:00 AM, Fromkin, Russ wrote: Vitali, Scott, Just wanted to check in to see how things are coming along with the updated system. Ben has provided some initial data that you have seen much better behaviors/performance, at least with the memory. Please let us know how things are coming along and if you run into any issues or have any questions. It would be good to hear how some of your other tests are comparing vs. the December work in OR and the more recent A0 remote access results. Thanks, Russ