PC Construction

It seems to me that this test is just a burn in on the processor. As the cores are all maxed out they are the bottleneck and you aren't testing the limits of memory speed (which you were talking about earlier) or other parts of the system. (I know you aren't bothered about graphics.)
Fine by me, I think my phrase "raw computing power" pretty much covers it, but I take your point about it not being a suitable comparison when running the RAM up to speed.
 
Last edited:
Re-running on my notebook - there was a noticeable change in fan speed when the workload started! Unfortunately this is an irresistible distraction from what I actually need to be doing...
 
Eek! Last time I looked, ZEN was on frame 70, then a few minutes later it had finished! The last 30 frames took less than 30 mins, the first 30 took nearly 6 hours. Stats coming when I can have my notebook back...
 
This may not sound like much, but I've been used to my Celeron notebook:

I've ripped a CD to MP3 which is now playing through the graphics card via HDMI to the TV, installed Virtualbox (involving a fair bit of head scratching before I managed to collect together all the components - why isn't there one "install this" button?), started off a Win 7 installation in a VM, and while all that is running also copied the Mandelbrot image set to a USB drive... all of which would have taken a considerable period on the old machine!

I must say I quite like Mint - it's sufficiently similar to legacy Windows that I can find my way around - and in some ways easier.

:):):)
 
Results from my Mandelbrot benchmarking, which are slightly disappointing in that ZEN is outperforming my Celeron T3100 notebook by a factor of only 10. Nonetheless, this still means that the full run (which took 12h48m with ZEN) would have taken over 5 days on my notebook.

Mandelbrot_Benchmark.png
 
why isn't there one "install this" button?
I've never had an issue, although I don't rely on the default repo version preferring to use the most recent version on the Virtualbox website. Or are you referring to having to also install the extension pack which is supplied separate because it's supplied under a different (non-GPL) licence (still trivial to do, mind)?

As for benchmarking, Michael Larabel that runs Phoronix authored an extensible benchmarking suite which he uses heavily for comparison purposes on Phoronix that can upload data for comparison on Openbenchmarking.org. Sounds like something you might care to look into?

Edit: For clarity, that is to say you can directly compare apples with apples using a standardised set of benchmarks
 
Last edited:
Not like you to take such a casual attitude towards this stuff.
It's not like I won't - it's just not at the top of the list yet. "Mirror" is sufficient summary of a future intention, and "does it matter?" simply means I don't want to be pinned down yet.

Phoronix had come onto my radar.

Anomaly: with an Ethernet cable plugged in (other end connects to my three-way HomePlug), the system won't start up from the power button (actually: screwdriver across the motherboard power button pins). Unplug the Ethernet, start it up, then reconnect the Ethernet, and it's fine. "Feels like" and earth loop issue.

Plays havoc with TV reception - not on the local TV/PVRs (which have the first point of call on the cable run) but further down the chain.

Virtualisation: I may have been naive about what this was going to achieve for me. My expectation was that the VM host would provide a simulated set of PC "hardware" with little exposure of the actual hardware it was running on, so that a Windows installation would not get upset even if it was run (as a VM) on an entirely different set of hardware. But looking at the system from within Windows, it is identifying the processor as a Ryzen 5 - so presumably exporting the VM to a different processor would still trigger the Windows allergic response. That's not what I wanted.
 
Virtualisation: I may have been naive about what this was going to achieve for me. My expectation was that the VM host would provide a simulated set of PC "hardware" with little exposure of the actual hardware it was running on, so that a Windows installation would not get upset even if it was run (as a VM) on an entirely different set of hardware. But looking at the system from within Windows, it is identifying the processor as a Ryzen 5 - so presumably exporting the VM to a different processor would still trigger the Windows allergic response. That's not what I wanted.
So what you want is CPU emulation but that will be at the cost of a big performance hit.
 
I don't see that it would be that much of a performance loss - yes, obviously it could not leverage any special acceleration features of a specific chip that have to be programmed, but aren't modern processors all basically backwards compatible in terms of the binary execution set (to the first 64-bit core, anyway)? Don't they all run the same code (minus special additions) but internally use look-ahead and speculative execution etc invisibly? Code can't "see" what happens inside a processor, it can only "know" what data gets set in memory and registers - that should be the same, regardless of the processor (it just gets there faster if the chip is faster or cleverer).

If there was a performance loss, it might not matter (being used to Windows on a 1.9GHz Celeron T3100).

Maybe there are settings in Virtualbox (or other VM host) that configure the reported processor. Maybe I need to use 32-bit to get this (not requiring a BIOS setting to turn on hardware virtualisation support).
 
Anomaly: with an Ethernet cable plugged in (other end connects to my three-way HomePlug), the system won't start up from the power button (actually: screwdriver across the motherboard power button pins). Unplug the Ethernet, start it up, then reconnect the Ethernet, and it's fine. "Feels like" and earth loop issue.
This may be a coincidence. I had trouble starting it this morning even without the Ethernet cable (or HDMI). I was beginning to think it was terminal until it started on the umpteenth attempt.
 
I think you have re-created a Humax box!

Do you have hdmi out on this machine? What's all this talk of VGA cables about? (Sorry, haven't read in detail.)
 
Anomaly: with an Ethernet cable plugged in (other end connects to my three-way HomePlug), the system won't start up from the power button (actually: screwdriver across the motherboard power button pins). Unplug the Ethernet, start it up, then reconnect the Ethernet, and it's fine. "Feels like" and earth loop issue.
Is it sitting on an anti-static bag or other form of insulation? I'm just wondering if any part of the MoBo is shorting underneath. Just to be sure I'd also try swapping the ethernet lead as well. Certainly this isn't right.

Plays havoc with TV reception - not on the local TV/PVRs (which have the first point of call on the cable run) but further down the chain.
Might be my sleep starvation, but clarification is required: are you referring to it's relative position from the master socket on a power-strip relative to other devices (power related) or RF interference? If the latter I wouldn't be surprised as after all it should really be in a case (Faraday cage).

Virtualisation: I may have been naive about what this was going to achieve for me. My expectation was that the VM host would provide a simulated set of PC "hardware" with little exposure of the actual hardware it was running on, so that a Windows installation would not get upset even if it was run (as a VM) on an entirely different set of hardware. But looking at the system from within Windows, it is identifying the processor as a Ryzen 5 - so presumably exporting the VM to a different processor would still trigger the Windows allergic response. That's not what I wanted.
No, you have the measure of it otherwise what would be the point of virtualising the O/S if not to make it portable across different platforms? At the risk of dishing out your own advice, I'd take the time to read the manual to familiarise yourself with how Virtualbox works. Specifically "1.13. Cloning virtual machines", "1.14. Importing and exporting virtual machines" and possibly "7.2. Teleporting" (although the latter is more oriented toward data centres where a live running virtual machine is duplicated over the network). You can also fluff around with adding (and on Linux subtacting) CPU's on the fly if you wish, although I expect this lies outside what you hope to achieve. The good news is you can always stick Virtualbox on your old machine and test the various duplication methods to ensure they work as you would expect.
 
Maybe there are settings in Virtualbox (or other VM host) that configure the reported processor. Maybe I need to use 32-bit to get this (not requiring a BIOS setting to turn on hardware virtualisation support).
Using Virtualbox, as long as the host you move the image to runs an Intel processor (i386/amd64 using Linux architecture nomenclature) you should be good to go.
 
This may be a coincidence. I had trouble starting it this morning even without the Ethernet cable (or HDMI). I was beginning to think it was terminal until it started on the umpteenth attempt.
Personally I'd want to be certain I didn't have a dud MoBo or power supply. If you're absolutely certain there is no chance of any MoBo shortage as you have it mounted currently, can't you rig up something a little less Heath Robinson than shorting the pins with a screwdriver? Also have you tried experimenting with "cold starting" using the master on/off switch on the PSU itself on your issues? Obviously there is a residual power supply to the board all the time when the PSU is on even if the machine itself isn't running.
 
Is it sitting on an anti-static bag or other form of insulation?
Answering myself, I've just checked the photo you posted. Whilst I can see it's sitting on an anti-static bag, it's still on a non-conductive surface (cardboard box). That's fine for build/testing of course but I'm wondering if running it for extended periods is building a static charge especially with the CPU fan whizzing around. The Motherboard itself would usually be grounded in several places when mounted in a chassis as well as through it's power connector. Maybe you could try earthing the metal cowl around the Ethernet slot?
 
Do you have hdmi out on this machine? What's all this talk of VGA cables about? (Sorry, haven't read in detail.)
I have a cast-off (but decent) monitor to use which is only VGA. I don't expect to have ZEN connected to my main sitting room TV forever! Cable should arrive tomorrow (the PS/2 keyboard + mouse to USB adapter arrived today, so I can plug in my wireless desktop and play from the sofa!).

Is it sitting on an anti-static bag or other form of insulation? I'm just wondering if any part of the MoBo is shorting underneath. Just to be sure I'd also try swapping the ethernet lead as well. Certainly this isn't right.
I'll keep an eye on it, but I need to ration my play time because of "stuff". The anti-stating bag should be static dissipative rather than conductive, so it should be OK (but a few stand-offs would be a good idea, and what I intended in the first place).

Might be my sleep starvation, but clarification is required: are you referring to it's relative position from the master socket on a power-strip relative to other devices (power related) or RF interference? If the latter I wouldn't be surprised as after all it should really be in a case (Faraday cage).
RF - yes, I was just commenting. I don't understand the fashion for replacing nice screening metal cases (against interference in both directions) with glass panels.
 
Back
Top