Jump to content
Forum upgrade is live! Read more... ×
Sign in to follow this  
g__day

iNeed to upgrade - which Intel socket 2066 motherboard?

Recommended Posts

Just imagine days when CPU cores are light switched - ten times the speed of electron travel, minimial heat multi cores galore. Expect to see it in our lifetimes!

Edited by g__day
  • Like 1

Share this post


Link to post
Share on other sites

it seems the settings used can have a huge effect on how multithreaded the software is

registering images can be almost completely single threaded or change a few settings and it takes twice as long due to computing the master files for dark\flat ect. which would be done in the blink of the eye with some settings but can take a minute or two each with modified settings which is a process that can make use of multiple cores

 

last night i downloaded these images to use

The Orion Nebula
34 Lights
15 Darks
25 Flats
25 Dark Flats

http://www.rawastrodata.com/dso.php?type=nebulae&id=m42

 

the feeling im getting from using this software at the moment is that extra cores will bring less than half the improvement they would bring when rendering a video

 

 

Edit:

streamed a video of settings and cpu usage

 

registering

https://www.twitch.tv/videos/200558794

stacking and adjusting image

https://www.twitch.tv/videos/200562763

Edited by Dasa
  • Like 2

Share this post


Link to post
Share on other sites

So I just tried once again for fun... on NGC 5137 - with 16 shots ranging in duration from 60 seconds to about 10 minutes. This subject has about 17,000 stars in it - so its a headache for DSS. DSS took about an hour simply to register all the star fields - then of course my PC crashed before it stacked the shots. My core temps were up around 71 degrees = wow! CPU usage even with a deep virus scan running in the background was only 70%??? and I/O wasn't that high either??? so this is only 160MB of data to process - I didn't include any darks, bias, flat files etc - just to keep it simple as possible.

 

Warrants more investigation later today!

Share this post


Link to post
Share on other sites

the feeling im getting from using this software at the moment is that extra cores will bring less than half the improvement they would bring when rendering a video

 

 

Edit:

streamed a video of settings and cpu usage

 

registering

https://www.twitch.tv/videos/200558794

stacking and adjusting image

https://www.twitch.tv/videos/200562763

I wonder if we could get an idea of how much effect HT has on it, whether we're actually getting +50% compute perf? Might be worth a try with HT disabled?

 

You can also definitely see the I/O bottleneck there... the cores waiting for the disk to load the files. NVMe RAID would be SWEeT!

Share this post


Link to post
Share on other sites

tried disable ht it made no difference if anything it was quicker with it off

dss is not actually io limited just switching between single threaded work load and multithreaded work load

i setup a ram drive and it didn't help and if you look at the disk usage in the video other than a few seconds when first opening and saving the files there isn't much in it so a single nvme drive should be sweet

 

so dss is very different in its requirements to the pixinsight benchmark

Share this post


Link to post
Share on other sites

Guys Hot damn those videos were great to watch - thank you! Never realised that Registering appears more single threaded and stacking multi threaded. ALso seeing you play with the Luminance levels after stacking and how fast it redrew was impressive to watch - that is at least twice as fast as mine!

 

M42 - Oriin Nebulae had about I think while I was watching the registering process about 120 stars detected - so a global cluster like Omega Cluster - where stars number in the high thousands up to 15,000 will be a pain methinks!

 

Interesting to watch memory, CPU amd I/O - that was really a treat thanks so much Dasa! I have put a request in to joing the Deep Sky Stacker group on Facebook - I want to ask them some questions on speed of processing tips and PC gear recommended!

Share this post


Link to post
Share on other sites

ALso seeing you play with the Luminance levels after stacking and how fast it redrew was impressive to watch - that is at least twice as fast as mine!

seems that is mostly single threaded ipc with which my mild 4.6ghz oc will be helping a bit

the fastest cpu for that would be a 8600k\8700k overclocked to ~5ghz

but obviously a cpu with more cores will complete the stacking quicker

unfortunately it would be a bit of a weak point for threadripper

 

you have me tempted to step outside with my sisters d750 and snap a few shots of the stars

she has some nice very nice fixed lenses from wide angle up to a big old 600mm zoom and being in a remote location its 15k to the nearest small town emitting any light pollution

Edited by Dasa
  • Like 2

Share this post


Link to post
Share on other sites

Oh that could be brilliant - dark skies to die for!

 

If you even did 15 - 30 second shots on a tripod there would not be noticeable star trails!

Share this post


Link to post
Share on other sites

 

ALso seeing you play with the Luminance levels after stacking and how fast it redrew was impressive to watch - that is at least twice as fast as mine!

seems that is mostly single threaded ipc with which my mild 4.6ghz oc will be helping a bit

the fastest cpu for that would be a 8600k\8700k overclocked to ~5ghz

but obviously a cpu with more cores will complete the stacking quicker

unfortunately it would be a bit of a weak point for threadripper

 

you have me tempted to step outside with my sisters d750 and snap a few shots of the stars

she has some nice very nice fixed lenses from wide angle up to a big old 600mm zoom and being in a remote location its 15k to the nearest small town emitting any light pollution

 

Someone buy the 8600K on PCCG before it sells!

 

EDIT: Great work Dasa

Edited by SceptreCore

Share this post


Link to post
Share on other sites

That certainly isn't a bad build spec - when you downgrade the CPU - the motherboard gets considerably cheaper - you can get by with a smaller PSU, cooler, case and retain much of what I want to do. Equally moving up from the 7820 to the I9 - costs you $500 for two extra cores, extra PCIE lanes and cache. Not really justified in my case. I will probably only add in 2-3 SSDs and 3-4 HDDs once I confirm whether one or more of them are failing.

 

Which leads me to - is there any really good software to tell if a Seagate HDD is occassionally prone to fail?

Share this post


Link to post
Share on other sites

That certainly isn't a bad build spec - when you downgrade the CPU - the motherboard gets considerably cheaper - you can get by with a smaller PSU, cooler, case and retain much of what I want to do. Equally moving up from the 7820 to the I9 - costs you $500 for two extra cores, extra PCIE lanes and cache. Not really justified in my case. I will probably only add in 2-3 SSDs and 3-4 HDDs once I confirm whether one or more of them are failing.

 

Which leads me to - is there any really good software to tell if a Seagate HDD is occassionally prone to fail?

Well I certainly wouldn't choose that myself... but I know that you like a flashy Gigabyte motherboard so I chose that... but you can definitely go something cheaper. You could even downgrade to a 1080 or 1070 Ti if you don't need the maximum of framerates. You don't even need RGB RAM... but they look so pretty.

Share this post


Link to post
Share on other sites

your original plan for 7820x may not have been a bad one really it hits a decent middle ground for dss with two extra cores and slightly slower game performance and lower max oc

7820X $700
Asus TUF X299 Mark 2 $340

CORSAIR Vengeance LPX 32GB (4 x 8GB) CMK32GX4M4E4133C19R $614 <newegg
Gigabyte nVidia GeForce GTX 1080 Ti Aorus OC 11GB $956
Samsung 960 PRO 512GB M.2 $319.2
HGST Ultrastar He8 HUH728080ALN600 8TB $492 <newegg
Corsair HX Series Modular HX750 750W 80 Plus Platinum $175 <ample power
Fractal Design Define R5 $155 < better airflow & quieter
Noctua NH-D15 $114 <quieter
$3865

Share this post


Link to post
Share on other sites

your original plan for 7820x may not have been a bad one really it hits a decent middle ground for dss with two extra cores and slightly slower game performance and lower max oc

7820X $700

Asus TUF X299 Mark 2 $340

CORSAIR Vengeance LPX 32GB (4 x 8GB) CMK32GX4M4E4133C19R $614 <newegg

Gigabyte nVidia GeForce GTX 1080 Ti Aorus OC 11GB $956

Samsung 960 PRO 512GB M.2 $319.2

HGST Ultrastar He8 HUH728080ALN600 8TB $492 <newegg

Corsair HX Series Modular HX750 750W 80 Plus Platinum $175 <ample power

Fractal Design Define R5 $155 < better airflow & quieter

Noctua NH-D15 $114 <quieter

$3865

Yes but it's poor value compare to compared to Ryzen. Save $500 and get the R7 1700?

 

Share this post


Link to post
Share on other sites

very true

7820x oc 4.8GHz vs 1700 oc 4GHz both with 3200 ram

gaming performance about the same

single threaded photoshop performance 7820x is ~15% faster

rendering video 7820x is ~25% quicker

 

so although the value isn't great there is more performance there

threadripper for a little more will eat it in the multithreaded performance but performance about the same as the 1700 in single threaded

Edited by Dasa

Share this post


Link to post
Share on other sites

The advantage in photo rendering tests are even less than the video ones, and that's likely the workload that'll be done.

 

Coffee Lake is out of the question. If the 8 core was out.. which is what the Z390 board is going to support, that might be different... but would it still use the ring bus architecture or go to mesh like Skylake X and lose its edge? Regardless it's still not here.

 

7820 or 1700 would be good buys and arguments for both. But for me, -50% cost for the 1700. it's not that far behind in the rendering workloads, and there's an upgrade path for Ryzen+

 

But if g_day didn't want to swap the CPU out and sell the old one later... then 7820 would be the go.

 

 

EDIT: In my silly opinion of course.

Edited by SceptreCore
  • Like 1

Share this post


Link to post
Share on other sites

Kind of an interesting analysis in several ways - especially what it says about me!

 

  1. Although in the past I have switched between AMD and Intel as it suits my gaming needs - when it comes to non gaming I seem to lean to Intel - not sure why my bias
  2. Although I have forked out maybe close to $30K for all my astronomy gear - building a rig to better handle astronomy workloads has me trying to allocate under $6K to a decent, future expandable rig - where as maybe $5K would be ideal for AMD and $7K for Intel. So ignoring value and looking at fit for purpose can't say why I didn't simply go for a 12 or 14 core I9 today for an extra $1K to $1.5K and just say hang the expense. Most likely because it's too much cash that can go elsewhere and I simply don't do astrophotography often enough to warrant the financial commitment of a bigger machine today.

I was really interested to see how well older CPUs heavily overclocked do in games. It makes me wonder how much of that is games not yet fully optimised for multi core CPUs. The Intel path means my budget is gouged on CPU and Motherboard pricing versus AMD - if folk can justify that - you can expand the system you want fairly easily. Wonder in a years time what CPUs will look like?

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
Sign in to follow this  

×