Author Topic: Observations on intel vs AMD  (Read 2033 times)

Doug Clark

  • Sr. Member
  • ****
  • Posts: 349
  • Karma: +11/-1
    • View Profile
Observations on intel vs AMD
« on: November 08, 2024, 03:24:30 pm »
I am current running AOS 5.1 on three machines:

Lenovo T530 laptop with an 2.8-GHz dual-core Intel Core i5-3360M (circa late 2012) with SATA SSD

HP EliteDesk 800 G4 65 watt  mini desktop with an Intel i5 8500 (circa early 2018) with very fast NVMe SSDs,

A desktop Ryzen 5 5600x on MSI PROB500-VC (circa late 2020) with very fast NVMe SSD and spinning hard drive.  The Ryzen 5 5600x and MSI PROB550-VC can still be purchased new as of this writing (Nov 2024). 

I offer my very unscientific observations in hope that anyone considering a machine replacement for AOS may find them useful.  My observations center on two areas: performance and compatibility - although my own personal priorities are compatibility and then performance.

Performance
-----------------
Graphical: I am not gaming on any machine so for me graphical performance on all three are basically the same. The machines are running both panorama (T530) and SNAP in both accelerated (Ryzen) and non-accelerated mode (EliteDesk 800), and for me there is no noticeable difference in graphical performance.  This includes text mode scrolling in editors and command line prompts, as well as graphical performance in VLC while playing movies.

Boot up time: Boot time, from power on to desktop ready is.  T530 - 31 seconds, EliteDesk - 31 seconds, Ryzen 5 - 72 seconds.  All machines are running MBR disks and all go through Airboot - so the times include me hitting Enter as fast as possible in Airboot.  All are running CSM/legacy mode.  All machines have UEFI bios.  The relative differences between the machines stays the same when re-booting, i.e. a warm boot.  The longer boot time for the Ryzen is almost all in the "what the hell are you doing in there" bios setup time, that is the time between turning on the machine and getting the Airboot prompt. For a warm re-boot on the Ryzen there is a multi-second delay between clicking the Shut Down menu item in the Desktop pop-up menu and the display of the extended shutdown prompt.  I have to point out that the Ryzen is running xWorkplace and the two Intel machines are running AOS Enhanced Desktop.

Desktop restart time:  Here I am timing how long it takes for the Restart Desktop prompt to show after clicking on the Restart Desktop prompt in the Desktop popup menu. T530 and EliteDesk are essentially instantaneous, Ryzen 5 is 7 seconds.  The actual desktop restart process, once you hit OK at the Restart Desktop prompt is essentially the same.

The bottom line is the T530 with a SATA SSD and slowest CPU boots as fast as the EliteDesk with the fast NVMe SSD, and twice as fast as the Ryzen with the fastest NVMe SSD I could buy plugged in to PCIe gen 4 socket.  For boot times it appears Intel has a significant edge.

General Performance:  It is hard to directly compare the performance of the machines because they are used for different tasks.  The EliteDesk is used as a file and database server (WSeB). The Ryzen and T530 are used as general purpose machines and do a lot of compling.
 
Once I (finally) get the Ryzen booted it is the fastest of the three machines, which is mainly noticeable in application startup time and compile times.  I suspect this is mostly due to the speed of drive access.  The Ryzen is running PCIe gen 4 with a very fast NVMe SSD.  The EliteDesk is running PCIe gen 3 with the fastest NVMe SSDs I could buy.  The T530 is running a SATA SSD.   

Compatibility
------------------
I have significantly more money invested in software then I ever will have in hardware, so compatibility is very important to me.  I run, on a regular basis,  both OS2 and Win-OS/2 software (both Win16 and Win32s), some of which was released in 1992.  Software compatibility is generally not a problem with OS/2 software.  I have a couple of applications from 1992 that run on OS/2 - they have a few issues but run.  The compatibility issues mostly arise in Win-OS/2 and here things get complicated.

The T530 is more compatible with Win-OS/2 software than either the EliteDesk or the Ryzen - in most ways.  It will run applications in seamless mode that will not run in seamless mode on the Ryzen or the EliteDesk;. the EliteDesk will not run Win-OS/2 or DOS in anything other than full screen mode.

On the other hand, I can install some Win-OS/2 applications on the Ryzen that won't install on the other two machines - even though the Ryzen has problems running some applications in seamless mode.

The bottom line is that for compatibility  it appears there is more of a difference in how old the machine is than there is between Intel and AMD processors.  With older generally being better.

Versions of AOS
-----------------------
The T530 will run earlier versions of AOS than 5.1, but Win-OS/2 will not run on that machine until AOS version 5.1.  You cannot install any version earlier than AOS v 5.1 on the Ryzen - the installer will not work.  You can install an earlier version of AOS on a different machine and move the drive over to the Ryzen but USB,  and perhaps networking, will not work.

I didn't try to install an earlier version than 5.1 on the EliteDesk but I suspect an earlier version would not install.

This is not unique to AOS.  By way of comparison you cannot install Win7 on the EliteDesk without heroic efforts and strange incantations.  The fact that you can install and run AOS on the latest hardware is truly an accomplishment.

Hope this is helpful to someone.

Eugene Tucker

  • Sr. Member
  • ****
  • Posts: 403
  • Karma: +15/-0
    • View Profile
Re: Observations on intel vs AMD
« Reply #1 on: November 08, 2024, 03:30:46 pm »
Than you for this report.

Roderick Klein

  • Hero Member
  • *****
  • Posts: 697
  • Karma: +14/-0
    • View Profile
Re: Observations on intel vs AMD
« Reply #2 on: November 08, 2024, 04:10:11 pm »
I am current running AOS 5.1 on three machines:

Lenovo T530 laptop with an 2.8-GHz dual-core Intel Core i5-3360M (circa late 2012) with SATA SSD

HP EliteDesk 800 G4 65 watt  mini desktop with an Intel i5 8500 (circa early 2018) with very fast NVMe SSDs,

A desktop Ryzen 5 5600x on MSI PROB500-VC (circa late 2020) with very fast NVMe SSD and spinning hard drive.  The Ryzen 5 5600x and MSI PROB550-VC can still be purchased new as of this writing (Nov 2024). 

I offer my very unscientific observations in hope that anyone considering a machine replacement for AOS may find them useful.  My observations center on two areas: performance and compatibility - although my own personal priorities are compatibility and then performance.

Performance
-----------------
Graphical: I am not gaming on any machine so for me graphical performance on all three are basically the same. The machines are running both panorama (T530) and SNAP in both accelerated (Ryzen) and non-accelerated mode (EliteDesk 800), and for me there is no noticeable difference in graphical performance.  This includes text mode scrolling in editors and command line prompts, as well as graphical performance in VLC while playing movies.

Boot up time: Boot time, from power on to desktop ready is.  T530 - 31 seconds, EliteDesk - 31 seconds, Ryzen 5 - 72 seconds.  All machines are running MBR disks and all go through Airboot - so the times include me hitting Enter as fast as possible in Airboot.  All are running CSM/legacy mode.  All machines have UEFI bios.  The relative differences between the machines stays the same when re-booting, i.e. a warm boot.  The longer boot time for the Ryzen is almost all in the "what the hell are you doing in there" bios setup time, that is the time between turning on the machine and getting the Airboot prompt. For a warm re-boot on the Ryzen there is a multi-second delay between clicking the Shut Down menu item in the Desktop pop-up menu and the display of the extended shutdown prompt.  I have to point out that the Ryzen is running xWorkplace and the two Intel machines are running AOS Enhanced Desktop.

Desktop restart time:  Here I am timing how long it takes for the Restart Desktop prompt to show after clicking on the Restart Desktop prompt in the Desktop popup menu. T530 and EliteDesk are essentially instantaneous, Ryzen 5 is 7 seconds.  The actual desktop restart process, once you hit OK at the Restart Desktop prompt is essentially the same.

The bottom line is the T530 with a SATA SSD and slowest CPU boots as fast as the EliteDesk with the fast NVMe SSD, and twice as fast as the Ryzen with the fastest NVMe SSD I could buy plugged in to PCIe gen 4 socket.  For boot times it appears Intel has a significant edge.

General Performance:  It is hard to directly compare the performance of the machines because they are used for different tasks.  The EliteDesk is used as a file and database server (WSeB). The Ryzen and T530 are used as general purpose machines and do a lot of compling.
 
Once I (finally) get the Ryzen booted it is the fastest of the three machines, which is mainly noticeable in application startup time and compile times.  I suspect this is mostly due to the speed of drive access.  The Ryzen is running PCIe gen 4 with a very fast NVMe SSD.  The EliteDesk is running PCIe gen 3 with the fastest NVMe SSDs I could buy.  The T530 is running a SATA SSD.   

Compatibility
------------------
I have significantly more money invested in software then I ever will have in hardware, so compatibility is very important to me.  I run, on a regular basis,  both OS2 and Win-OS/2 software (both Win16 and Win32s), some of which was released in 1992.  Software compatibility is generally not a problem with OS/2 software.  I have a couple of applications from 1992 that run on OS/2 - they have a few issues but run.  The compatibility issues mostly arise in Win-OS/2 and here things get complicated.

The T530 is more compatible with Win-OS/2 software than either the EliteDesk or the Ryzen - in most ways.  It will run applications in seamless mode that will not run in seamless mode on the Ryzen or the EliteDesk;. the EliteDesk will not run Win-OS/2 or DOS in anything other than full screen mode.

On the other hand, I can install some Win-OS/2 applications on the Ryzen that won't install on the other two machines - even though the Ryzen has problems running some applications in seamless mode.

The bottom line is that for compatibility  it appears there is more of a difference in how old the machine is than there is between Intel and AMD processors.  With older generally being better.

Versions of AOS
-----------------------
The T530 will run earlier versions of AOS than 5.1, but Win-OS/2 will not run on that machine until AOS version 5.1.  You cannot install any version earlier than AOS v 5.1 on the Ryzen - the installer will not work.  You can install an earlier version of AOS on a different machine and move the drive over to the Ryzen but USB,  and perhaps networking, will not work.

I didn't try to install an earlier version than 5.1 on the EliteDesk but I suspect an earlier version would not install.

This is not unique to AOS.  By way of comparison you cannot install Win7 on the EliteDesk without heroic efforts and strange incantations.  The fact that you can install and run AOS on the latest hardware is truly an accomplishment.

Hope this is helpful to someone.


Couple of notes. The difference between speed of UEFI and MBR boot should be just about 0. The difference between how ArcaOS in UEFI mode AND MBR starts might down to less then maybe half a second difference's if not less.

Its not down to CPU support for DOS and WIN/OS2 support. What is the most likely cause is the speed of the system and possibly, memory layout.

If you have an issue with WIN/OS2 programs and it can be reproduced then open a ticket at https://mantis.arcanoae.com.
You can also try run the system with the line psd=acpi.psd /maxcpu=1 in the config.sys. This will run with one core.

Some installers can not handle volumes for ArcaOS bigger then 2 GB and might hang.
So that are some things you can try to pinpoint issue's.

Based on how I understand the video subsystem in OS/2 works for WIN/OS2 this should just works and this mostly is routed via GRADD (VMANWIN.SYS I think the driver is called).
But in a UEFI environment there MIGHT be unknown bugs.  Every bug report you could report to Arca Noae can help improve the code.
But its hard to debug VDM (DOS/WIN/OS2 code).

Roderick

Doug Clark

  • Sr. Member
  • ****
  • Posts: 349
  • Karma: +11/-1
    • View Profile
Re: Observations on intel vs AMD
« Reply #3 on: November 08, 2024, 04:47:44 pm »
Roderick,

Thanks for the input.

I should have mentioned that  psd=acpi.psd /maxcpu=1  in CONFIG.SYS does fix the issues I have with Win-OS/2 software running on the Ryzen system, as does running those applications in full screen mode.  But running apps in seamless mode is pretty important to me. And I should make it clear that some Win-OS/2 applications run fine in seamless mode on the Ryzen machine. Some lock up the machine pretty quickly, some take a longer while to lock up the machine.  I do have a ticket open for this particular issue.

The title Intel vs AMD is somewhat misleading because I am not sure how much is just limited to the CPU and how much influence supporting chipsets, motherboard and bios have - but those other factors are usually closely related to the CPU selected.

Roderick Klein

  • Hero Member
  • *****
  • Posts: 697
  • Karma: +14/-0
    • View Profile
Re: Observations on intel vs AMD
« Reply #4 on: November 08, 2024, 07:39:38 pm »
Roderick,

Thanks for the input.

I should have mentioned that  psd=acpi.psd /maxcpu=1  in CONFIG.SYS does fix the issues I have with Win-OS/2 software running on the Ryzen system, as does running those applications in full screen mode.  But running apps in seamless mode is pretty important to me. And I should make it clear that some Win-OS/2 applications run fine in seamless mode on the Ryzen machine. Some lock up the machine pretty quickly, some take a longer while to lock up the machine.  I do have a ticket open for this particular issue.

The title Intel vs AMD is somewhat misleading because I am not sure how much is just limited to the CPU and how much influence supporting chipsets, motherboard and bios have - but those other factors are usually closely related to the CPU selected.

My understanding is not that great. But just like with OS/2 and Windows 3.1 the programs running are pretty agnostic when it comes to chipsets tec. It means they do not really talk to the BIOS or anything. This is not like a DOS application that talks to the BIOS.  That is why I keep leaning towards this is a possible timing issue somewhere.

Roderick

 

Dariusz Piatkowski

  • Hero Member
  • *****
  • Posts: 1355
  • Karma: +26/-0
    • View Profile
Re: Observations on intel vs AMD
« Reply #5 on: November 08, 2024, 08:01:16 pm »
Hey Doug!

...desktop Ryzen 5 5600x on MSI PROB500-VC (circa late 2020) with very fast NVMe SSD and spinning hard drive.  The Ryzen 5 5600x and MSI PROB550-VC can still be purchased new as of this writing (Nov 2024). 
...
Performance
-----------------
Graphical: I am not gaming on any machine so for me graphical performance on all three are basically the same. The machines are running both panorama (T530) and SNAP in both accelerated (Ryzen) and non-accelerated mode (EliteDesk 800), and for me there is no noticeable difference in graphical performance.  This includes text mode scrolling in editors and command line prompts, as well as graphical performance in VLC while playing movies...

I am very interested in the AMD Ryzen stuff, especially in light of this being a configuration that allows you to run SNAP.

I have an old ATI X850XT PE video card here that, until I decide to buy one of these curved high-resolution displays, will continue to serve me in a dual-head configuration.

OK, so what GPU are you running? ...as this would certain appear to be a viable approach.

Further on, can you post the matching SysBench results for your combo? (I've attached mine)

I recently upgraded (yes, LOL, I am dealing with some rather ancient hardware here by today's standards, but alas, where is the fun in all this otherwise?) from a AMD Phenom II X6 110T (6-core) to AMD FX-8370 Piledriver (8-core) CPU configuration.

Without getting into the nitty gritty CPU architecture details, I have been somewhat dissapointed with the outcome...but hey, only $50 later it can hardly be thought off as 'wasted spend'!

So bottom line here being: performance benchmarks aside, it is the whole feel of the machine that needs to be right, and before one spends the $$$ to go there I would ideally love to be able to do some kind of an assessment as to what the outcomes may be.

Doug Clark

  • Sr. Member
  • ****
  • Posts: 349
  • Karma: +11/-1
    • View Profile
Re: Observations on intel vs AMD
« Reply #6 on: November 09, 2024, 06:18:33 am »
Dariusz,

I have attached Sysbench results - it seems not much of my hardware was recognized so I edited some of the descriptions in the results.

My 2 top priorities for a machine are a big screen and a good keyboard.

For graphics I am running a Radeon X600 because SNAP supports the chipset and therefore I can run dual monitors - as shown in the attached picture desktop_2.jpg  Which gets me a big screen for a very reasonable amount of money. 27" 1920x1080 LCD monitors run about $100 each nowdays.

For keyboards I used a IBM model M for years because they typed so nicely.  I fact my daily driver until 2 weeks ago was a Model M made in 1993, and I bought a number of them to take to client sites. My new keyboard shown in the picture is essentially a Model M restomod.  It is made with entirely new Model F internals, in a steel case designed to look like the Model M.  It types even better than the original Model M.  If you have ever typed on an old Model F keyboard you know what I am talking about. It is put out by Model F Labs.

This combination has a great feel - both screen and input wise. But installing an older card like the X600, or the X850, can be a challenge on a new machine, because of hardware issues and the installer for AOS 5.1

The issues:

1) AOS 5.1 installer will not run unless the graphics card/chipset supports GOP (graphics output protocol), which apparently came along with UEFI.  The Radeon X600, and probably the X850 do not support GOP and therefor the AOS installer will not run with those cards.  I may be stating this incorrectly but that essentially is the issue.

2) I could not find a single Ryzen 5 processor (when I purchased) that had an on-board graphics processor and supported a PICe gen 4 sockets.  Apparently the chips had a limited number of gen 4 lanes and dedicated them to the on-board graphics chip.  I wanted at least one gen 4 socket so I could get a faster NVMe SSD.  I might have been able to get a Ryzen 7 that had the graphics processor and at least one gen4 NVMe socket but those generate more heat and didn't come with a cooler so you pay more intial costs and operating costs for probably not much if any performance gain.

3) I purchased a graphics card that would support GOP so the AOS installer would run, then took the card out and replaced with the Radeon X600 to get SNAP support and dual monitors. The cheapest I could  find was a Radoen HD 5450 put out by VisonTek that cost about $70.

4) This means I needed a maintenance partition on a hard drive because I would not be able to boot from the install USB stick when problems arose unless I switched video cards .

5) I chose to format the drives as MBR to avoid the performance penalty that comes with the GPT driver - David A. talks about this at the 2023 Warpstock in Tempe Az. during one of his presentations.  I had someone suggest that GPT allows you to customize block sizes on the drive and avoid some additional wear that happens on SSDs but I decided I cared more about the performance than the wear rate, and I wasn't smart enough to figure out how to do that anyway.

6) There is only one boot manager type application/utility that  (potentially) allows you start in UEFI mode then then allows you to boot to Legacy or CSM. mode. That utility is reFind.  But even though others could get that to work I could not.  But the documentation on the reFind site describing how machines decide to boot into either UEFI or Legacy and the problems with this is excellent.

7) The Radeon will boot in pure UEFI mode but the screen does strange things for about 15 seconds before starting the desktop normally.  However if you boot that way Win-OS/2 doesn't work.

8) This means you have to boot in Legacy or CSM mode which means AirBoot is your boot manager.   AirBoot has an issue, at least with my Ryzen machine, where it will lock up if you use a USB keyboard to make any selections in AirBoot.  I have a very small PS/2 type keyboard plugged into the PS/2 port of the motherboard just for navigating AirBoot. Once you get past AirBoot, and before you get to AirBoot, USB keyboards work just fine.  It is only in AirBoot that the problem exists. Well - if you have an error when booting and you are using  a USB keyboard you cannot Hit Enter to Continue. The PS/2 keyboard also works in those situations.

9) Finally - I have speed issues with RealTek network chips when using Netbios.  These issues do not show up when using TCP/IP.  Since my file server/NAS is WSeB it makes sense for me to be using Netbios on the  OS/2 machines.  This meant installing an older PCI Intel NIC card so that I could use Intel drivers and get decent network speed.  Which meant getting an PCI to PCIe adapter. The first one I tried did not work.  The second one made by StarTech did.  It is listed on the OS2WORLD wiki under the Hardware - SCSI section.

I was forced to upgrade my old AMD machine when the ASUS motherboard stopped working.  The transition  to a new machine was very painful and took weeks, but I am very happy with the end results.  It feels very good, and my graphics performance, while not as good in Sysbench as yours is,  feels very good to me.  For example I can drag application screens across the desktop with no noticeable delay.

I do have some issues with some Win-OS/2 applications but I think started having those issues when I upgraded AOS versions on my older AMD FX4350/ASUS machine and didn't realize what the cause was at the time.

BTW - I have an old Radeon X800 in the basement and might drag it out to see what it does for the graphics benchmarks.  I switched to the X600 because the X800 took up two slots, had a loud fan, and was a power piggy.  This new machine is essentially silent - I have to get my head right next the fans in the case to hear anything at all.  It is uses about 60 watts at the desktop at idle (without a screen saver) 90 watts when the screen saver kicks in.  At boot-up it hits about 120 watts briefly.

Hope this helps.
« Last Edit: November 09, 2024, 06:20:12 am by Doug Clark »

Roderick Klein

  • Hero Member
  • *****
  • Posts: 697
  • Karma: +14/-0
    • View Profile
Re: Observations on intel vs AMD
« Reply #7 on: November 09, 2024, 12:27:01 pm »
Hey Doug!

...desktop Ryzen 5 5600x on MSI PROB500-VC (circa late 2020) with very fast NVMe SSD and spinning hard drive.  The Ryzen 5 5600x and MSI PROB550-VC can still be purchased new as of this writing (Nov 2024). 
...
Performance
-----------------
Graphical: I am not gaming on any machine so for me graphical performance on all three are basically the same. The machines are running both panorama (T530) and SNAP in both accelerated (Ryzen) and non-accelerated mode (EliteDesk 800), and for me there is no noticeable difference in graphical performance.  This includes text mode scrolling in editors and command line prompts, as well as graphical performance in VLC while playing movies...

I am very interested in the AMD Ryzen stuff, especially in light of this being a configuration that allows you to run SNAP.

I have an old ATI X850XT PE video card here that, until I decide to buy one of these curved high-resolution displays, will continue to serve me in a dual-head configuration.

OK, so what GPU are you running? ...as this would certain appear to be a viable approach.

Further on, can you post the matching SysBench results for your combo? (I've attached mine)

I recently upgraded (yes, LOL, I am dealing with some rather ancient hardware here by today's standards, but alas, where is the fun in all this otherwise?) from a AMD Phenom II X6 110T (6-core) to AMD FX-8370 Piledriver (8-core) CPU configuration.

Without getting into the nitty gritty CPU architecture details, I have been somewhat dissapointed with the outcome...but hey, only $50 later it can hardly be thought off as 'wasted spend'!

So bottom line here being: performance benchmarks aside, it is the whole feel of the machine that needs to be right, and before one spends the $$$ to go there I would ideally love to be able to do some kind of an assessment as to what the outcomes may be.

Just some clarifications.

* UEFI GOP is an interface ArcaOS uses from the UEFI firmware.  So if a video card does not support UEFI GOP that is it.
Infact your BIOS might also not work (no screen output). I am not an expert at this.

* The loss of performance with the GPT filter driver is about 15%.  But with an SSD of NVMe storage you will not notice this.

* The usage of SNAP for multi screen support is the only way out. As you noticed. The SNAP and Panorama performance, you will not even notice it in practice.
Also SNAP is waaaay more complicated compared to Panorama. Panorama is a lightweight driver compared to SNAP.

Roderick

ivan

  • Hero Member
  • *****
  • Posts: 1603
  • Karma: +20/-0
    • View Profile
Re: Observations on intel vs AMD
« Reply #8 on: November 09, 2024, 01:48:18 pm »
Hi Doug

Out of curiosity what is your total resolution with that dual monitor setup?

I have a Ryzen 3 processor that drives my 30 inch 2560x1600 monitor without problems via a USB/HDMI 4 port switch so allowing my Linux Mint, Win data logger and Warp 4 boxes access as necessary.

Martin Iturbide

  • OS2World NewsMaster
  • Global Moderator
  • Hero Member
  • *****
  • Posts: 5026
  • Karma: +44/-1
  • Your Friend Wil Declares...
    • View Profile
    • Martin's Personal Blog
Re: Observations on intel vs AMD
« Reply #9 on: November 09, 2024, 02:25:03 pm »
Hello Doug

Sorry to drift away from the subject, but your setup picture looks great. I'm obsessed with your "IBM 6094-010 Dials" and your "IBM Lighted Program Function Keyboard". I don't know if they work on OS/2 or what would I do with those, but they look so damn cool !!!!.

Regards
Martin Iturbide
OS2World NewsMaster
... just share the dream.

Dariusz Piatkowski

  • Hero Member
  • *****
  • Posts: 1355
  • Karma: +26/-0
    • View Profile
Re: Observations on intel vs AMD
« Reply #10 on: November 09, 2024, 03:40:33 pm »
Hi Doug,

I have attached Sysbench results - it seems not much of my hardware was recognized so I edited some of the descriptions in the results.

My 2 top priorities for a machine are a big screen and a good keyboard.

For graphics I am running a Radeon X600 because SNAP supports the chipset and therefore I can run dual monitors - as shown in the attached picture desktop_2.jpg  Which gets me a big screen for a very reasonable amount of money. 27" 1920x1080 LCD monitors run about $100 each nowdays.
...
This combination has a great feel - both screen and input wise. But installing an older card like the X600, or the X850, can be a challenge on a new machine, because of hardware issues and the installer for AOS 5.1...

Excellent, thank you for sharing.

Yeah, so it would seem that my focus and priority on having not just a usable but also 'fitting' OS/2 experience matches yours pretty closely. The extensive summary & feedback you are providing here is GOLDEN!!! I appreciate the time you have put into this...

...
1) AOS 5.1 installer will not run unless the graphics card/chipset supports GOP (graphics output protocol), which apparently came along with UEFI.  The Radeon X600, and probably the X850 do not support GOP and therefor the AOS installer will not run with those cards.  I may be stating this incorrectly but that essentially is the issue.

2) I could not find a single Ryzen 5 processor (when I purchased) that had an on-board graphics processor and supported a PICe gen 4 sockets.  Apparently the chips had a limited number of gen 4 lanes and dedicated them to the on-board graphics chip.  I wanted at least one gen 4 socket so I could get a faster NVMe SSD.  I might have been able to get a Ryzen 7 that had the graphics processor and at least one gen4 NVMe socket but those generate more heat and didn't come with a cooler so you pay more intial costs and operating costs for probably not much if any performance gain.

3) I purchased a graphics card that would support GOP so the AOS installer would run, then took the card out and replaced with the Radeon X600 to get SNAP support and dual monitors. The cheapest I could  find was a Radoen HD 5450 put out by VisonTek that cost about $70.

4) This means I needed a maintenance partition on a hard drive because I would not be able to boot from the install USB stick when problems arose unless I switched video cards .
...

So this has been quite a concern of mine: how the heck to transition to the new hardware while not having to give up most of the functionality I enjoy today. Great insights right there...especially the stuff dealing with the GPU swap as I wouldn't have known about it and as best as I can tell this isn't called out anywhere else (albeit I migth have of course missed it, if it was).

My son has left me his "old" Ryzen hardware for an upgrade consideration: Asus Prime X470-PRO, AMD Ryzen7 motherboard, Ryzen7 2700X CPU, GeForce GTX 1060 GPU, 16G RAM and a Samsung 970 Pro, M.2 NVMe PCI-E 512GB drive. I keep on trying to carve out the time to take on that projects, but all too often it is just sooo much easier to plug away on the stuff that's working for me right now! lol

...BTW - I have an old Radeon X800 in the basement and might drag it out to see what it does for the graphics benchmarks.  I switched to the X600 because the X800 took up two slots, had a loud fan, and was a power piggy.  This new machine is essentially silent - I have to get my head right next the fans in the case to hear anything at all.  It is uses about 60 watts at the desktop at idle (without a screen saver) 90 watts when the screen saver kicks in.  At boot-up it hits about 120 watts briefly...

I am thinking that the X600=>X800 move will just about close out whatever gaps you may be seeing right now.

The SNAP drivers could be made a tad better by providing additional controls for the GPU, but maybe architectually this wasn't quite fully completed even when the SNAP development was ongoing. Case in point, my X850XT PE is actually the R480 chip with 540 MHz core frequency and 590 MHz memory frequency, but SNAP only runs it at 400/400:

Code: [Select]
Graphics device configuration:
  Manufacturer......... ATI
  Chipset.............. Radeon X850 Series
  Bus Type............. PCI Express
  Memory............... 24576 KB
  DAC.................. ATI Internal 24 bit DAC
  Clock................ ATI Internal Clock
  Memory Clock......... 400 MHz
  Default Memory Clock. 400 MHz
  Maximum Memory Clock. 400 MHz
  Driver Revision...... 3.2, Build 29
  Driver Build......... May  2 2017
  Certified Version.... Not Certified

So a while back I wrote a little util to manipulate those settings thinking that if the SNAP drivers are detecting these and the SDK API is described as being able to SET these...I might as well maximize my hardware!

No-go, while the util runs and no errors are returned, the SysBench results do not change.

Anyways, no biggie, especially in light of the as-is doing exactly what I need it to do so far.

Doug Clark

  • Sr. Member
  • ****
  • Posts: 349
  • Karma: +11/-1
    • View Profile
Re: Observations on intel vs AMD
« Reply #11 on: November 09, 2024, 06:18:27 pm »
Roderick,

The MSI motherboard/BIOS will switch from UEFI/CMS mode to CMS/UEFI mode during boot if it does not detect GOP support in the graphics card, which pretty much insures that you see SOMETHING - at least with this motherboard/BIOS combination.  But the installer won't get the screen where you select system maintenance or install.  I submitted a ticket and some data to AOS to see it they would support the card in the installer and the answer was no - which I completely understand.

As you said, I can't detect a performance difference between Panorama and SNAP.

IVAN,
The resolution is 3840 x 1080 - exactly double the width of the standard 27 inch monitor's 1920 x 1080.  I wish I had more vertical resolution but it isn't worth the cost or hassle to switch - although in theory I could switch to the x800 and 34 inch monitors to get a 1600 vertical resolution.  The physical width (left to right) of the two 27 inch monitors is just under 49 inches.  For me that is about the maximum usable size, and that is with me sitting back a ways from the desk.

I didn't think Panorama would support a horizontal resolution of 3840 so I tried a 49 inch monitor with picture by picture in order to get rid of the black frame "bars" in the middle of the screen by switching to a single monitor.  The idea was I would drive each input of a 49"  monitor that supported picture by picture with a separate monitor output of SNAP.   It worked - but the result was a very large black area on the monitor between the left and right outputs, which was considerably wider that the physical frame that you can see between the left and right 27" monitors pushed together as I have them.  I had to use VGA to HDMI converters to get from the output of the X600 to the inputs of the Samsung 49 inch monitor, so I don't know if it was a monitor issue or convert issue.  But I found out that a 49" monitor is very heavy and wobbly on my desk.  So that experience made my happier with my two monitor setup.  Plus these two monitors are a LOT cheaper than a single 49 inch monitor.

I am using a IOGear 4-port DualView VGA KVMP switch and the monitors have VGA/HDMI inputs.  So I can switch both monitors at the same time. I have hooked up to the KVM  to OS/2, Win7 and a modified version of Win10 (Windows AME) that removes the telemetry and switches some of the built in Windows parts (like file explorer) to look and function like they did in Win7.  I am guessing that most OS/2 users who have a desktop PC also use some kind of KVM.

Martin,

The LPFK (lighted programmable function keyboard) and the 8 dial thing next to it are old products IBM sold mainly for use with RS600 machines running CAD software.  Each of the 8 dials has 256 positions as it rotates.  Each "key" in the LPFK emits a signal when pressed, and the LED for the key is separately addressable.  My plan for the LPFK is to have a WPS object/driver where you can drag and drop another object on the key to assign the key to some action for the object: Open, Run, Close etc.  The LED should be lit when the open is open or running, and off when the object closes or stops. I am waiting to finish up with a new version of my WinZombe program object because that can signal when an application closes or stops running - which means I would not have to poll the window list or active processes to find that out and turn off the light for a key.

I am also planning layers for the LPFK, which would function like layers in a normal programmable keyboard - each layer being a separate set of assignments for the keys and LEDs.  That way you could assign a layer to an application so the application could control the keys and LEDs in "its" layer either through a named pipe or by sending WPS setup strings to the LPFK.  A perfect application for that, in my opinion, is X10 home automation.  There is already an OS/2 X10 interface in the House/10 application and it has a Rexx interface.  The keys could then be assigned to separate devices in the house: lights, garage doors, sprinkler circuit switches, etc. and the dials could be assigned for sending dim/bright signals to a light.  In that case the LPFK would function like a desktop version of an X10 keypad - see attached pic.

You probably didn't notice the black Space Mouse on the desk.  This device essentially replaces the LPFK and the 6 of the dials on the IBM 6094-10 Dials device.

I am pretty much done with the WinZombe object except for the feature that Dariusz asked for a few years back - a folder containing recently opened documents.  The recently opened documents folder object is populated by the WinZombe object when it passes a document to an application - either by drag and drop or by association when a data file object is double clicked.  I have to finish building the part that deletes a shadow in the folder when the total exceeds a user specified number based on when the shadow was added to the folder.

Dariusz,

I am guessing the your son's "hand me up" system will work very well as an upgrade to what you are using today.  Good luck with that.


Remy

  • Hero Member
  • *****
  • Posts: 765
  • Karma: +12/-1
    • View Profile
Re: Observations on intel vs AMD
« Reply #12 on: November 10, 2024, 11:27:10 am »
note: I think https://www.oo-software.com/en/shutup10 would be safer than AME for win10/11

JTA

  • Jr. Member
  • **
  • Posts: 65
  • Karma: +0/-0
    • View Profile
Re: Observations on intel vs AMD
« Reply #13 on: November 10, 2024, 03:17:48 pm »
AME is a key component of my virtualization processes over in the virt sub-forum ... it makes for a perfect, stripped down HostOS, that then runs any other os in a VM, such as ArcaOS. Gets me x64 apps into ArcaOS.

  os2world.com/forum/index.php/topic,3498.0.html

I've been running AME for nearly two years now, and there hasn't been any problems, security issues, or such that I can find. On the other hand, there also have not been any data leaks going the other way back to MS, nor have I had to put up with their crud (changes, advertising, etc.) coming my way.

I can report that AME is good stuff ... saves me *hours* of trying to strip out by hand all of MS' garbage in their os's ...

Roderick Klein

  • Hero Member
  • *****
  • Posts: 697
  • Karma: +14/-0
    • View Profile
Re: Observations on intel vs AMD
« Reply #14 on: November 10, 2024, 09:17:10 pm »
Roderick,

The MSI motherboard/BIOS will switch from UEFI/CMS mode to CMS/UEFI mode during boot if it does not detect GOP support in the graphics card, which pretty much insures that you see SOMETHING - at least with this motherboard/BIOS combination.  But the installer won't get the screen where you select system maintenance or install.  I submitted a ticket and some data to AOS to see it they would support the card in the installer and the answer was no - which I completely understand.

As you said, I can't detect a performance difference between Panorama and SNAP.

So the BIOS is displayed on the screen ? What about the ArcaOS boot menu when you install from install media ? Is the ArcaOS boot logo displayed ?
You saying that just the ArcaOS installer is never installed ? Do you have text output during boot or also not ?

Roderick