Download: MSI Afterburner 4.5.0

JonasBeckman

2018-04-24 19:49

The added option to delete the cache should be fine I'm assuming. After upgrading to a new Windows 10 build RTSS tends to pop up a warning prompt (EDIT: First time it's run that is.) about the hook not initializing properly and a compare of the config file at this point will generally have the cache addresses as 000000 I believe, restarting the software usually corrects it and if not a reboot will but now with the option to delete and rebuild the cache since a few versions back that too should work. EDIT: This thread moves fast. 🙂 (EDIT: And some cumulative update changing the DirectX files will usually see the address modified but I've never gotten a warning prompt from RTSS and it just handles it, cleaning the cache after major updates couldn't hurt though.)

#5540538

Im_Special

2018-04-24 19:51

Unwinder:

No, it is not the case and you’re not supposed to do so when upgrading AB/RTSS to newer version. But it can be the case after major OS version upgrade, in case if RTSS injection cache gets damaged.

Alright, I'll keep that in mind for whenever MS releases 1803 to the public, if things do go weird. I don't mind doing "clean installs" its just that it requires a little more prep on my part first, remember settings and such.

#5540557

Unwinder

2018-04-24 20:54

JonasBeckman:

The added option to delete the cache should be fine I'm assuming.

Yep, resetting the cache via options will also do the trick. "Clean install" is just faster way to instruct a user.

JonasBeckman:

After upgrading to a new Windows 10 build RTSS tends to pop up a warning prompt (EDIT: First time it's run that is.) about the hook not initializing properly and a compare of the config file at this point will generally have the cache addresses as 000000 I believe, restarting the software usually corrects it and if not a reboot will but now with the option to delete and rebuild the cache since a few versions back that too should work. EDIT: This thread moves fast. 🙂 (EDIT: And some cumulative update changing the DirectX files will usually see the address modified but I've never gotten a warning prompt from RTSS and it just handles it, cleaning the cache after major updates couldn't hurt though.)

Those are two different cases and two different situations. The first one is the message you mentioned, it is displayed when RTSS is started into a sandbox environment for a first time and it see that some system software (e.g. antivirus) is trying to replace DirectX runtimes with proxy libraries on the fly, so offsets for injection cache cannot be detected reliably. Upgrading DirectX files to newer versions is a different case, it is normally handled transparently for end user and RTSS should redetect the offsets without displaying any notifications.

#5540558

Unwinder

2018-04-24 20:56

gydj:

So what's your suggestion that I can get it work? Right now it's like I adjust power limit to 50%, click apply, the slider jump back to 0. Same to the core and mem: I change the number, click apply, everything revert back to default. Appreciate your help!

Sorry, I gave up on "how do I configure AB" questions in the forum, probably somebody else can help you. I'm focused on development related things only.

#5540565

xxela

2018-04-24 21:21

[SPOILER][/SPOILER]

Unwinder:

Attached video is demonstrating specific process memory usage monitoring in details, step by step. The same approach can be used for other process specific counters like CPU usage, handle count etc.

I wached the video and I tried for process memory usage monitoring and it worked fine but when I chose "processor time" option for the specific process it show me only 100% [SPOILER]http://oi65.tinypic.com/2j1lso7.jpg[/SPOILER]

#5540567

Unwinder

2018-04-24 21:28

xxela:

I wached the video and I tried for process memory usage monitoring and it worked fine but when I chose "processor time" option for the specific process it show me only 100% http://oi65.tinypic.com/2j1lso7.jpg

You misunderstood interpretation of "% Processor time". It is documented here, for example: https://social.technet.microsoft.com/wiki/contents/articles/12984.understanding-processor-processor-time-and-process-processor-time.aspx It is a total CPU usage across all logical CPUs. So if you have let's say 8 logical CPUs then it is measured in 0-800% range, so to convert it to 0-100% you should define divider as 8. Also, don't forget to enable dynamic mode for this counter (it is disabled on your screenshot). Otherwise the graph won't appear if you start MSI AB without target process (bf1.exe) running in background (dynamic counters are not validated during creation).

#5540582

Koniakki

2018-04-24 22:19

I kinda miss the early days where Afterburner was been developed(I mean before it reached a stable/final state) and we or at least I, was waiting impatiently for every new revision/release! :P I say that with the utmost respect! Thank you Unwinder!

#5540617

ruthan

2018-04-25 01:44

Did some testing, not all processes are listed or could be monitored in PerfMon - for example not works on Kingdom come (i tried write procces name and pid too)- and GUI, need adjustment to be more fail proof - empty input boxes are fail - there should be comboboxes with predefined options..

#5540623

Thunk_It

2018-04-25 02:36

Many thanks to Guru3D, Hilbert, and the innovative Unwinder. I really appreciate these ongoing Afterburner updates!

#5540636

chispy

2018-04-25 04:34

Thank you so much for the new release Unwinder. ( Downloading Now ) Kind Regards: Chispy

#5540638

GSDragoon

2018-04-25 04:42

Thank you for the update!

#5540639

Rmosher

2018-04-25 04:52

Anyone else having trouble getting the new metrics for 1%, 0.1% and min/max/average frame rates to show in the osd? For the life of me I can't seem to get them to work...

#5540642

Unwinder

2018-04-25 05:12

ruthan:

Did some testing, not all processes are listed or could be monitored in PerfMon - for example not works on Kingdom come (i tried write procces name and pid too)- and GUI, need adjustment to be more fail proof - empty input boxes are fail - there should be comboboxes with predefined options..

PerfCounter plugin is fully open source, you're free to locate and fix the problem if you beleive that not all processes are being listed or empty input boxes are "fail".

#5540644

Unwinder

2018-04-25 05:16

Rmosher:

Anyone else having trouble getting the new metrics for 1%, 0.1% and min/max/average frame rates to show in the osd? For the life of me I can't seem to get them to work...

It is not new for this version and it was repeated a few times in the forum that fps min/avg/max/1/0.1 statistics are available in benchmark mode only. Start a benchmark with assiciated hotkey.

#5540645

Rmosher

2018-04-25 05:40

Unwinder:

It is not new for this version and it was repeated a few times in the forum that fps min/avg/max/1/0.1 statistics are available in benchmark mode only. Start a benchmark with assiciated hotkey.

Ah I see, thanks for the reply. Have a good one

#5540652

Unwinder

2018-04-25 07:33

FAQU:

Hey guys ! Is there any way to display GPU's power usage as an average ? For example I'd like to see the power usage like 131 W rather than the current 130.7 W

That's not called averaging, you're talking about rounding and using different output format. Currently output format override is supported for plugin-specific hardware monitoring data sources only, built-in sources always output data as is. So if you're displaying power usage via any plugin you can change format to "%.0f".

#5540680

reflex75

2018-04-25 09:44

First thank you very much for this new version of such a great tool! But I have to report something weird regarding CPU cores temperatures. I have a 7700k with 4 cores and I have noticed a big difference in temperature with one the cores compared to another great tool such as HWINFO. I did a screenshot bellow where you can see the difference in temperature for the third core (I removed the other 4 threads temperatures reports because they have the same value as 4 cores) Here you can check the "CPU5" in Afterburner is 5° colder (77°) than the same "Core#2" in HWinfo (82°) : https://image.noelshack.com/fichiers/2018/17/3/1524641380-afterbunrer-cores-temp-diff.png Which one is correct? It looks to me that HWInfo is more accurate, but can't prove it. Something else weird in Afterburner is cores 5 and 7 temperatures are linked together but in the opposite way: when one has a pic then the other has the same pic but in the other way...

#5540684

Unwinder

2018-04-25 09:55

reflex75:

Which one is correct? It looks to me that HWInfo is more accurate, but can't prove it.

It is rather hard (if not say impossible) to misread DTS just for one thread, but I cannot prove it too. CPU temperature monitoring implementation in CPU.dll plugin is also open source and included in SDK, so you may try to find the reason there.

#5540697

Rx4speed

2018-04-25 11:15

Does this still not control all three fans of the FTW3 separately?

#5540715

Unwinder

2018-04-25 12:33

Why still and why would it magically start controlling it? It is proprietary and undocumented EVGA's PWM controller, how do you imagine to see support for that in MSI AB? Primary GPU fan controllers for both NVIDIA and AMD cards are supported and directly controlled inside NV/AMD drivers, all critical safety measures as fan stop control, access synchronization for multiple software tools etc are implemented there. EVGA decided that they don't need all that and used additional own controller, invisible to standard NVIDIA fan control software infrastructure. Which means that any software needs to access undocumented fan controller directly at low level in order to work with additional fans. That's not something you can expect to see either in AB or in any other NVIDIA software fan controller - too much development cost to pay because of one vendor's exclusivity tricks.