There is a new program that shows you the IPC of your CPU.

D

Dannotech

n00b
Joined
Dec 10, 2016
Messages
42
Hope you will head over to my website and check it out!
http://www.dannotech.com/statuscore


(Sorry guys, it's not compatible with Windows 7 yet since I haven't had much luck debugging the Windows 7 driver but I'm still working at it!)

(Also, the temperature is not working for AMD based processors because I haven't done the thermal status registers yet.. Working on that, too!)



statuscore-app.png
 
FX 8300 is read as the correct cpu. However, the speed is way too low and the temp is not showing correctly. (It only shows 4 cores and the temp shows 100C all the time.) Thanks for the effort though.

Edit: Reading is fundamental, I just saw that the temp is not there yet for AMD processors, oops. :D
 
running 6 cores on each keeps it above 3Ghz/core, 7 & 8 drops down to 2.8 and 2.7Ghz/core.
statuscore-12 cores.jpg
statuscore-16 cores.jpg


AMD doesn't read correctly,
statuscore-AMD-FX-8150-8-Core.jpg
statuscore-AMD-FX-8150-8-Core-Stressed.jpg
 
Last edited:
dook43 said:
Source available on github? Is your code signed?
Click to expand...

Yes the code is signed; no the source is not open. Also I'm mostly focused on getting it ready for Ryzen so older AMD processors will have to wait a bit long. The main problem is like I said the thermal reading isn't programmed yet and the bigger issue is how Windows reports the CPU to my program as being hyperthreaded even though it is not. Basically to work around this problem I need to write my own CPUID routine that can distinguish between CMT and SMT since Windows doesn't do it for me.

I have preordered a Ryzen 7 1800X and I plan on having it compatible on Day 1. Hopefully.. =)
 
Ahh crap Microsoft why make it so hard? I suspect this is the problem:


Driver Signing Policy
Last Updated: 1/24/2017

Starting with new installations of Windows 10, version 1607, Windows will not load any new kernel mode drivers which are not signed by the Dev Portal. To get your driver signed, follow these steps:

  1. Get an EV Code Signing Certificate. All drivers submitted to the portal must be signed by an EV certificate.
  2. Submit your new driver to the Windows Hardware Developer Center Dashboard portal.

So I already paid $500 for the EV code certificate-- now I gotta sign up for WHQL certification? </sigh>
 
Yeah I seem to be having trouble getting the driver installed. It doesn't show up in the device manager after I try to install it.
 
Dannotech said:
Yes the code is signed; no the source is not open. Also I'm mostly focused on getting it ready for Ryzen so older AMD processors will have to wait a bit long. The main problem is like I said the thermal reading isn't programmed yet and the bigger issue is how Windows reports the CPU to my program as being hyperthreaded even though it is not. Basically to work around this problem I need to write my own CPUID routine that can distinguish between CMT and SMT since Windows doesn't do it for me.

I have preordered a Ryzen 7 1800X and I plan on having it compatible on Day 1. Hopefully.. =)
Click to expand...

any particular reason why its closed-source? not a fan of random stuff that installs drivers that I can't see :p
 
*raises hand*
Please, for the love of god, can you at least provide a download link for a version that doesn't come with VC++ packages? I don't want to have to download 25MB every time there's an update. :) I have limited bandwidth (and for most of the day a slow connection to boot), so 25MB adds up. I'm cool with it being a text link, buried at the end of the page for that matter heh

I'll let you know if it works next time I restart this comp. Due to the aforementioned bandwidth/slow connection, I don't update this system, so I had no issues installing the driver :p
 
Dannotech said:
Hope you will head over to my website and check it out!
Click to expand...

That's an impressive program Dannotech.

Here's the results with my i5 at idle and with 1 and 4 cores stressed:
32965582492_43bca1f38e_o.jpg


For those of you like myself who had trouble trying to get it to run under windows 10, after you follow the installation instructions in the readme.txt file, create a shortcut to the Statuscore.exe file in the x64 directory.
 
Seems like I need to redo my thermal paste now. This was after 20 minutes. Although watching coretemp while gaming I never see temps this high. I tried getting my buddy with a Phenom II 965 BE to run it, but it shows his CPU temp as 100 all the time and is unable to run it.

statuscore.jpg

statuscore2.jpg

Voltage was set to auto, manually set to 1.32 and upped it 100mhz. Lower temps now.

xCQrn



l5GerF9
l5GerF9

cZ5xC
 
Last edited:
Isn't ipc is something that changes with each icache data / dcache address stream? What's the instruction mix?
 
Nuby1Canuby said:
Has anybody run Statuscore on their Ryzen yet? I'm curious what the MIPS will be.
Click to expand...


So I'm working on Ryzen compatibility. Currently it only detects 8 cores and no threads so the IPC will be totally incorrect. So it runs, but the results are not valid.
 
Dannotech said:
So I'm working on Ryzen compatibility. Currently it only detects 8 cores and no threads so the IPC will be totally incorrect. So it runs, but the results are not valid.
Click to expand...
Still wish you'd offer a package w/o VC 2015 Redist for those of us who already have downloaded it. Had to install the driver through Dev Mngr though, as right-clicking and Install wasn't actually making it show up in Dev Mngr.

Anyways. Your March 15th update indeed sees all 16 threads on mine. HOWever, there's a peculiarity. It isn't treating the "HT" in an unbiased manner it seems? That, or Thread2 of each Core really is lower performance, even when no load is put upon it. Which is odd based on what I was told on how these threads function.

Either way, here are the results, run under Balanced plan for Boost (HiPerf did not impart any positive change):

16 Thread:
Statuscore W10Px64 R7 1700X@Stock 16T.png


8 Thread - Odd Cores:
8T Evens.png


8 Thread - Even Cores:
8T Odds.png


2 Thread:
2T.png


1 Thread:
1T.png



EDIT: Just text errors on my part.
 
Last edited:
Greetings,

I've been trying to get this working w/out success. Installation was fine and the thing opens. The issue is that it other than the cpu id string and temps, nothing else is populated. 0MHz, no activity, no speed listing for each core, no IPC, no MIPS.

Observe the attached file. Note the driver is installed, and I've even rebooted in an attempt to determine if that might be the issue. No joy. This is version 17.04.24/x64. I could not find a contact email or a any generic way of providing feedback. A google search turned up this thread. If you actually _want_ feedback you would do well to go get a throw-away gmail account and stuff that into the help menu _OR_ a reference on your web page. Having to back-channel track you down is ass.

Please advise.

statuscore_issue.png
.
 
PrimalNaCl said:
Greetings,

I've been trying to get this working w/out success. Installation was fine and the thing opens. The issue is that it other than the cpu id string and temps, nothing else is populated. 0MHz, no activity, no speed listing for each core, no IPC, no MIPS.

Observe the attached file. Note the driver is installed, and I've even rebooted in an attempt to determine if that might be the issue. No joy. This is version 17.04.24/x64. I could not find a contact email or a any generic way of providing feedback. A google search turned up this thread. If you actually _want_ feedback you would do well to go get a throw-away gmail account and stuff that into the help menu _OR_ a reference on your web page. Having to back-channel track you down is ass.

Please advise.

View attachment 25356 .
Click to expand...


Are you running it inside a Virtual Machine? I don't think the VM can provide the counters since they aren't virtualized-- in theory it could report the global counters but I think they just opt to return 0. I haven't tried it myself. Thanks for the feedback, I'll update my home page with an e-mail address. That was a dumb thing to forget =)
 
Formula.350 said:
Anyways. Your March 15th update indeed sees all 16 threads on mine. HOWever, there's a peculiarity. It isn't treating the "HT" in an unbiased manner it seems? That, or Thread2 of each Core really is lower performance, even when no load is put upon it. Which is odd based on what I was told on how these threads function.
Click to expand...

The reason the IPC is different between thread1 and thread2 on the core is because the stress test itself is asymmetrical; thread 0 is mostly ALU while thread 1 is ALU and memory transactions. If both programs are running the exact same program then it doesn't get quite as high combined IPC because there is resource contention. This was by design and confirmed that it does indeed draw more power at the wall for a more, uh, we'll say 'rigorous' stress test. Thanks for the feedback!
 
Dannotech said:
The reason the IPC is different between thread1 and thread2 on the core is because the stress test itself is asymmetrical; thread 0 is mostly ALU while thread 1 is ALU and memory transactions. If both programs are running the exact same program then it doesn't get quite as high combined IPC because there is resource contention. This was by design and confirmed that it does indeed draw more power at the wall for a more, uh, we'll say 'rigorous' stress test. Thanks for the feedback!
Click to expand...
Gotcha :) Thank you for clarifying!
 
You must log in or register to reply here.
Back
Top