$1000 USD CAD and Rendering Workhorse. Getting the Balance right?

[mnementh]

Not at all, but since the OP is clearly on what I would consider a tight budget, for his limited amount of money that extra $40-80 could mean getting something else that is far more useful to them.

Also, just because it's the latest and greatest doesn't mean you should adopt it the first chance you get.

A-Bit brought out the first ATA 66 motherboards, which had a fatal flaw that randomly corrupted your HDDs making the bus unusable.
Fijitsu brought out the first budget home 6-10GB HDDs, that every single one failed due to the new method of encapsulating the controller IC.
Intel mass produced and sold the Intel Atom CPUs to the enterprise sector for mission critical infrastructure where flip-chip BGA construction was used, which are now all failing due to unforeseen issues with the at the time new technology.
Samsung brought out the first 1TB home SATA SSDs that suffered a fatal performance flaw due to issues with the wear levelling implemented in silicon that was rectified in later models.
AMD brought out the Ryzen 7 series of CPUs that have a critical bug that exhibits under Linux when doing multi threaded workloads causing a full system halt that was fixed in later revisions.

These are just a few examples of new tech having critical bugs/flaws in new unproven technology that has bitten the early adopters.

Remember those first motherboards with USB 3.0, which used a separate controller chip which was often actually slower and less reliable than the native USB 2.X connections on the same boards? Or the same story when SATA3 was introduced? Lots of fun, lots of confused people. I'm not saying that'll be the case here, but they're definitely examples of early adopter woes.

Now there's something we can agree upon. I've been bitten by the early-adopter snake; it lies in wait for all of us tech-heads.  But truly; rarely has the cost of that mistake been so small as seen here. 

I know bean he can handle that little bit of extra dosh; and IF he gets a raging nerd-boner for something he WILL spend the money. I also know a little more about his projected workload. He really does want an 8K video rendering beast. We've already been over all that in thread.

And since we're getting all historical here... Historically speaking and technologically as well, we are literally on the verge right now of a generational leap in hardware, not an incremental one. A shifting point where what used to be top-of-the-line quickly becomes entry-level, if relevant at all. The timing is right, all the indicators point to it... it appears the problem for a lot of people is that THIS TIME, it's AMD pushing the envelope. 

I think for that price it's a small enough gamble, and the rewards already visible enough that it's better to not wait for Intel to get around to it.

mnem
boo!

[gnif]

Air is quieter then AIO, but custom loop is even quieter and more flexible IMO Here is my current radiator.



However it's now under a shroud and the pump is outside too. I love having my TR 1950X clocked at 4GHz, c-states disabled, and zero room noise, and with the current outside temperature of 1.2C

[Mr. Scram]

Air is quieter then AIO, but custom loop is even quieter and more flexible IMO Here is my current radiator.

(Attachment Link)

However it's now under a shroud and the pump it outside too. I love having my TR 1950X clocked at 4GHz, c-states disabled, and zero room noise, and with the current outside temperature of 1.2C

(Attachment Link)

That's awesome! Is that a full copper setup? I've always toyed with the idea of doing something similar, but I'm absolutely positively too lazy to maintain a custom loop consistently over the lifetime of a machine. Forgetting to dust your air cooler won't get you into trouble as quickly and is more easily fixed. Dead reliability combined with low maintenance is generally what I'm aiming for.

[mnementh]

I don't know how many actually tried a water-cooling and what have changed for last 5-6 years with tech,
but I were trying to "eliminate" a noise for non-overlocking X-series and Xeon class CPUs that sat in boxes near to me.

This is my one of worst purchase in decade, obsolete sh$t with very creepy noise!

Eventually, money very well spend on a bigger case and traditional "beefy" coolers.

I am. I'm using the cheapest 120x240 AIO I could buy, BOUGHT ON PURPOSE to see how horrible it could be. Turns out, not at all. Fair moldings, glas-reinforced plastic, solid copper plate with proper milling and micro-grooving on the wet side. I can see it turning in my AMD dashboard, and pump noise is nonexistent. The VRM switching noise from my MB is louder. Six 120mm speed-controlled fans also make a whisper of noise that covers all other noise; but seriously... with the fans turned off, all I can hear is the whistle from my VRM.

The REASON I went liquid is BECAUSE I saw it deployed in commercial settings in my work environment. I figured if they're using it on servers, we must have most of the bugs worked out by now. My machine runs pretty much non-stop if I'm home; I stream Amazon music whether or not I'm not rotting my brain in threads like this one. 30-45° idle to full workload at 22-25° ambient, baby. And easy as pie to mount.

And I expect current offings from Corsair, etc to be markedly better quality yet; so yes... I believe liquid-cooling is finally a mature enough technology to be considered "mainstream".

mnem
*tzzzzt*

[gnif]

I agree entirely that AMD are the ones pushing the envelope however I am also well aware (because I am working with AMD on them) that the new AMD generations have major bugs that AMD had fixed that keep coming back in each generation.

AMD Threadripper:
  * Issue that prevents PCI devices from being reset making them inaccessible until a hard reboot is performed.
  * CPU scheduler on the last core was not exposed to the OS preventing it from being set to a performance profile.
  * CPU core numbering didn't follow the already set Intel standard preventing the OS from knowing which threads were paired for each HT pair per core.
  * CPU core numbering was changed to fix the above, but then reverted in the next update, and then fixed again in the update after that
  * Missing CPU scheduler bug returned and then was fixed again
  * NUMA issues... oh so many NUMA issues
  * IOMMU groupings keep getting changed around every few updates (if you use SR-IOV or VFIO, you care about this)

These are all mostly fixed now except for the reset issue, it still plagues some as it's partly CPU related and chip-set related.
Then AMD released the Zen+ CPUs a year later and PCI reset issues that were fixed for TR are back, and the cause is the same issue...

AMD Vega (All variants):
  * Random hard crashes under Linux with the official AMDGPU drivers
  * Broken support to reset the SOC and return it to an operational state
  * Missing support to program the SOC for PCIe 3.0 bus speeds under Linux (maybe windows too)

If you ever wanted to use NPT (nested page tables) for virtualization with sr-iov devices (ie, network/gpu device pass-through) your performance would have been sub-par because of a bug that AMD failed to identify and fix in the Linux kernel for 10+ years which I ended up fixing.

This makes me doubt the new PCIe 4.0 technology and chipsets, I personally will be holding off until the tech has seen more widespread usage.

[mnementh]

Air is quieter then AIO, but custom loop is even quieter and more flexible IMO Here is my current radiator:

   However it's now under a shroud and the pump is outside too. I love having my TR 1950X clocked at 4GHz, c-states disabled, and zero room noise, and with the current outside temperature of 1.2C

Now THAT is an unfair advantage, and hardly representative!!!

But seriously... AIOs are now being designed and targeted to the "quiet performance" market. See where the $75 Corsair H100b I suggested comes up both on cooling and noise compared to most other coolers... and you KNOW it will maintain that cooling more consistently and with a flatter response slope (until it dies, of course): https://www.anandtech.com/bench/CPUCooling/774

I think the big argument here is not over the tech itself... it's over an unreasonable expectation of lifetime. Yes, it is unreasonable to expect an AIO to last more than 3-5 years. That's fine, as long as you treat it as a consumable supply like a fan. Yes, SOME fans last 10 years or more... I have a Silent Copper cooler that is 15 years old and was still working, still quiet on my old build just before I went with the AIO.

But EXPECTING that lifespan out of a cooler is unreasonable. And poo-pooing liquid-cooling by comparison to that as others in here have done is unreasonable.

mnem
 

[gnif]

That's awesome! Is that a full copper setup? I've always toyed with the idea of doing something similar, but I'm absolutely positively too lazy to maintain a custom loop consistently over the lifetime of a machine. Forgetting to dust your air cooler won't get you into trouble as quickly and is more easily fixed. Dead reliability combined with low maintenance is generally what I'm aiming for.

Thanks, yes it's full copper if you count the brass radiator as copper . If you're aware of galvanic corrosion it's easily preventable allowing you to mix metals in your loop, just make sure there is no electrical path between the components of differing metals outside of the coolant itself. I have run copper blocks with aluminium radiators for years without issue simply by mounting the radiator using nylon screws.

Having the radiator outside makes maintenance so much easier, I soldered on a tap to cut the supply so I can simply go outside and turn the tap off, then disconnect the loop letting the just the pipes drain out so I can work on the PC. That said though I am yet to have to actually service this setup, the thermal mass is just insane and with the surface area I don't even need to run fans, so dust buildup is not an issue.

[Mr. Scram]

Now THAT is an unfair advantage, and hardly representative!!!

But seriously... AIOs are now being designed and targeted to the "quiet performance" market. See where the $75 Corsair H100b I suggested comes up both on cooling and noise compared to most other coolers... and you KNOW it will maintain that cooling more consistently and with a flatter response slope (until it dies, of course): https://www.anandtech.com/bench/CPUCooling/774

I think the big argument here is not over the tech itself... it's over an unreasonable expectation of lifetime. Yes, it is unreasonable to expect an AIO to last more than 3-5 years. That's fine, as long as you treat it as a consumable supply like a fan. Yes, SOME fans last 10 years or more... I have a Silent Copper cooler that is 15 years old and was still working, still quiet on my old build just before I went with the AIO.

But EXPECTING that lifespan out of a cooler is unreasonable. And poo-pooing liquid-cooling by comparison to that as others in here have done is unreasonable.

mnem
 

If I've learnt anything it's that the silent computing crowd is anything but uniform. One guy calls his system dead silent, the other can't stand to bear it. There's a huge product and end consumer gradient.

[Simon]

Having the radiator outside makes maintenance so much easier, I soldered on a tap to cut the supply so I can simply go outside and turn the tap off, then disconnect the loop letting the just the pipes drain out so I can work on the PC. That said though I am yet to have to actually service this setup, the thermal mass is just insane and with the surface area I don't even need to run fans, so dust buildup is not an issue.

If I ever did water cooling I would do that. I work for a radiator company and it was instantly obvious to me that these vehicle radiators that dissipate KW's of heat will take the sub 0.1KW output of a CPU in their stride, yes masses of surface area and masses of inertia.

I once poured a kettle of boiling water into a radiator about 200x100x30mm, it didn't even warm up (I was hoping to clean it out)

[Simon]

If I've learnt anything it's that the silent computing crowd is anything but uniform. One guy calls his system dead silent, the other can't stand to bear it. There's a huge product and end consumer gradient.

People have different tolerances to noise and noise pitch. I find fans highly annoying and am the one at work that cleans the air con filters out in the hope of cutting the noise levels. I am air cooled on my PC but an idling Ryzen 7 2700 uses next to nothing and I have lots of fans but set to very slow. I don't think fan noise in proportional to speed, it is exponential so a fan at 20% is virtually silent.

[gnif]

If I've learnt anything it's that the silent computing crowd is anything but uniform. One guy calls his system dead silent, the other can't stand to bear it. There's a huge product and end consumer gradient.

Yes, I couldn't agree more, I thought my older R7 system was quiet on it's custom water loop until I started to want to perform voice recordings in my "quiet" office. I quickly found out how "not quiet" my office was. Each source of noise I eliminated just exposed another source of noise. In the end I put the radiator and pump outside, put waterblocks on my GPUs, modified my 2RU UPS (24V fan on 12V) and then found it was still too loud so moved it out of the room, configured my SAS array to power down while not in use (only used during a daily backup for 30 minutes).

Each stage of making things more quiet, I became more attune to noises in the room that used to be drowned out by the last thing that was noisy. I finally got to a point where the only noise I can hear is the very slight hiss from my power amp when I leave it turned on.

Now when you enter my office it sounds like there is nothing running at all, and it's actually a bit eerie.

[mnementh]

I agree entirely that AMD are the ones pushing the envelope however I am also well aware (because I am working with AMD on them) that the new AMD generations have major bugs that AMD had fixed that keep coming back in each generation.

AMD Threadripper:
  * Issue that prevents PCI devices from being reset making them inaccessible until a hard reboot is performed.
  * CPU scheduler on the last core was not exposed to the OS preventing it from being set to a performance profile.
  * CPU core numbering didn't follow the already set Intel standard preventing the OS from knowing which threads were paired for each HT pair per core.
  * CPU core numbering was changed to fix the above, but then reverted in the next update, and then fixed again in the update after that
  * Missing CPU scheduler bug returned and then was fixed again
  * NUMA issues... oh so many NUMA issues
  * IOMMU groupings keep getting changed around every few updates (if you use SR-IOV or VFIO, you care about this)

These are all mostly fixed now except for the reset issue, it still plagues some as it's partly CPU related and chip-set related.
Then AMD released the Zen+ CPUs a year later and PCI reset issues that were fixed for TR are back, and the cause is the same issue...

AMD Vega (All variants):
  * Random hard crashes under Linux with the official AMDGPU drivers
  * Broken support to reset the SOC and return it to an operational state
  * Missing support to program the SOC for PCIe 3.0 bus speeds under Linux (maybe windows too)

If you ever wanted to use NPT (nested page tables) for virtualization with sr-iov devices (ie, network/gpu device pass-through) your performance would have been sub-par because of a bug that AMD failed to identify and fix in the Linux kernel for 10+ years which I ended up fixing.

This makes me doubt the new PCIe 4.0 technology and chipsets, I personally will be holding off until the tech has seen more widespread usage.

Yes, but THOSE KINDS of problems ARE NOT AMD-exclusive.  Intel and their chipsets have been fraught with boneheaded mistakes all along as well; and lets not forget we have two whole generations of computers out there... nearly every personal-use machine in the wild... that is just a massive security risk (and producing flawed results) due to speculative execution as it is implemented today. This is all a direct result of the dangerously close relationship between Intel and Microsoft. Even the related problems that AMD has with speculative processing, while generally not as severe, are STILL a result of having to work with the product of that unholy cabal. 

And in AMD's defense... a lot of your miseries with AMD and *NIX are by definition self-inflicted. It's not like there isn't a whole raft of similar types of problems... a flotilla of them, even... with Intel and *NIX. 

Be glad for AMD and them being willing to make mistakes to push the envelope. If not for them and Intel had their way, all "personal computers" would still be 32-bit, and everybody waiting, waiting, waiting with baited breath every quarter as they dole out a few hundred more MHz on the umpteenth revision of the original PENTIUM platform.

mnem
 

[gnif]

  You entirely missed the point, pushing the envelope doesn't mean they are bad or should be avoided, it does however means that they are prone to making mistakes just like intel and every other company on the face of the planet which is why I don't advocate being an early adopter of this technology, not because of hearsay but personal experience with this new frontier which I avidly support.

I have issues with *NIX and AMD are because I am a software engineer and I am helping to fix them to make AMD a better platform which is why I have a direct line to engineers at AMD and even had the fortune to communicate directly with Lisa Su on Vega reset issues. These issues also DO affect Windows, but because of the closed source nature of Windows and how the task scheduler works we just get a random "windows update" that makes AMD perform better with no real details. I can say for a fact that it's the same issues we are seeing in Linux

The difference is, we have an update that fixes these issues a full year before Windows.

If not for them and Intel had their way, all "personal computers" would still be 32-bit, and everybody waiting, waiting, waiting with baited breath every quarter as they dole out a few hundred more MHz on the umpteenth revision of the original PENTIUM platform.

Please fact check before you make such assertions, 64-bit was available to the home computing group well before the Athlon 64 was released. While in the end it was a commercial flop, the Itanium (2001) was available well before the x86_64 (2003) architecture to the home market (remember XP 64-bit), and even earlier was the IBM PowerPC POWER3 (1998), or even the Sun UltraSPARC from 1995. While you may not consider these home PCs, the equivalent PC enthusiasts of that era either had, or had access to these machines. In fact I have an UltraSPARC in my garage. You can even go as far back as 1991 with the MIPS R4000 which was used in the 1996 Nintendo 64

64-bit was the natural progression, not because Intel or AMD had to say so, but because we had hit the 3.2GB RAM limit (pre-PAE) and in order to make use of more ram a solution was required, the most obvious being to move to a 64-bit architecture while retaining backwards comparability with x86 (which is why the Itanium was a flop in the home market)

[mnementh]

Yeah, but somebody's gotta be the first. I like shiny-new bleeding edge EVERY ONCE IN A WHILE. I didn't miss the point; I was coming to the same conclusion from a different route. 

And YES, Itanium was available...  for an ungodly amount, if you wanted to run NT or some other server/workstation OS. I wonder why it was a commercial flop. It was still not even close to being properly supported by Microsoft/Windows who deliberately castrated XP 64-bit so it WOULDN'T run worth a shit until Intel got their asses in gear. It's not hearsay to me; I lived through it. And I still have the XP/64 licenses to prove it.

And yes, you'd think THAT experience would have taught me not to be an early adopter... 

Having the radiator outside makes maintenance so much easier, I soldered on a tap to cut the supply so I can simply go outside and turn the tap off, then disconnect the loop letting the just the pipes drain out so I can work on the PC. That said though I am yet to have to actually service this setup, the thermal mass is just insane and with the surface area I don't even need to run fans, so dust buildup is not an issue.

If I ever did water cooling I would do that. I work for a radiator company and it was instantly obvious to me that these vehicle radiators that dissipate KW's of heat will take the sub 0.1KW output of a CPU in their stride, yes masses of surface area and masses of inertia.

I once poured a kettle of boiling water into a radiator about 200x100x30mm, it didn't even warm up (I was hoping to clean it out)

My first foray into liquid-cooling decades ago on an AthlonIIx64 was with a leftover NOS heater core for a Ford Pickup truck. It was ~ 200mm x 200mm x 50mm, and it was easy to solder 3/8" hose barbs on the tubes. I attached it and a 120VAC water-fountain pump to a small 120 VAC tabletop box fan for a quick & dirty setup that actually worked remarkably well, but you needed to turn fan/rad and all sideways to get it to burp whenever you moved the thing. Compared to the screaming 60mm CPU fans of the day, that thing was effing silent.

mnem
Don't read this.

[gnif]

And YES, Itanium was available...  for an ungodly amount, if you wanted to run NT or some other server/workstation OS. I wonder why it was a commercial flop. It was still not even close to being properly supported by Microsoft/Windows who deliberately castrated XP 64-bit so it WOULDN'T run worth a shit until Intel got their asses in gear. It's not hearsay to me; I lived through it.

Microsoft had nothing to do with the flop of the Itanium, the lack of backwards compatibility was the issue and even Intel directly acknowledged it.

The Itanium was a bad move both from a software and hardware point of view, every vendor had to redevelop their software and/or drivers to make it operate on this new architecture practically overnight. It took Microsoft several years to update windows for x86_64 and the result was Vista, was this also Microsoft trying to "deliberately castrate" their own software?

For years many people had 64-bit capable CPUs (AMD and Intel) but were not running them in 64-bit mode simply because OS support wasn't great yet and vendors knew it, so they didn't provide it, CPU vendors even went as far as to add PAE to allow addressing more then 3.2GB of RAM because of how slow 64-bit software adoption was. I don't think you understand how complex the paradigm shift to a 64-bit architecture truly was and why early attempts such as XP 64bit didn't meet expectations.

In short, to say that we only have 64-bit due to AMD is simply not true, if it wasn't for the entire industry pushing for it, and the lessons learned by Motorola, IBM, Intel and AMD we would likely still be using 32-bit.

[mnementh]

Well, sure, before it was JUST AMD pushing them we had Motorola, CYRIX & TI too.  The point is that Intel showed the world exactly how they expected the CPU lifecycle to play out with the Pentium platform. A few hundred MHz on the same design for quarter after quarter... play out the design for every cent they can get, rather than the constant drive we have now towards better and faster based on what technology can actually do, NOT what they can market. All they want is for the world to go back to those halcyon days of yore... 

While saying "If AMD hadn't released a 64-bit processor aimed at the Windows-user market with a competitive pricing scheme, consumers would STILL TODAY be relegated to 32-bit with 64-bit reserved only for servers and high-end workstations." is a bit of an exaggeration, it is not MUCH of one. At the very least it would have been years or even a decade later; when Intel decided they'd milked 32-bit for all it was worth.

And none of this changes the fact that Intel has a dangerously close relationship with Microsoft, just as Microsoft also has a dangerously close relationship with Dell, both of which gain market share simply because of those preferential relationships.

Perhaps my view of this history we've both lived through is cynical; but I feel it is more a matter of not allowing the power players any more slack than absolutely necessary. I know they're in it to stack things in their favor; that is the nature of things. I don't confuse "choosing the least evil of the available evils" with "endorsing or actually liking" them.

But I still like everything I've seen about the X570 product line, and I'm still gonna buy it and build around it. 

mnem

[beanflying]

To early for popcorn so I opted for a two coffee read 

Seems the firmware on the 570 boards is still about 'beta' based on some follow up talking head youtube videos not that I am going there. My $ are going elsewhere where they will do some good with the system NOW not what might be later. All the planned bits will work on the 550/570 platform should I think it needed or a benefit later. Going AIR cooling for much the same reasons.

More importantly the Guatemalan HueHueTenango is a bit fresh yet but shows promise.

[beanflying]

Apart from the CPU and GPU the $ have been dropped. $528 USD ($830 AUpesos including GST)

Corsair Vengeance LPX 32GB C16
Gigabyte B450 WiFi MB (ATX not mITX)
Samsung 970 Plus 500GB
H500 Cooler Master Case
650 Bronze semi modular PS Cooler Master

All items were purchased from Bricks and Mortar Australian businesses with Ebay stores. Interesting I went to the trouble of looking at Amazon, New Egg and evilbay along with the sellers www websites - Evilbay with discounts won by a chunk.

Another coffee then time to find a 3700X 

[Simon]

I still don't get anyones need for 32GB of RAM unless they have an actual use for it. I do everything on 16GB and no pagefile.

[Black Phoenix]

 

[beanflying]

I still don't get anyones need for 32GB of RAM unless they have an actual use for it. I do everything on 16GB and no pagefile.

Not sure if you have had a complete read of the thread but DaVinci Resolve and 4K video is the culprit in my case. 16GB would work absolutely but 32GB will do it better and faster.

[Black Phoenix]

Davinci is way well optimised and supported compared with Premiere. More and more people are changing to Davinci instead of Premiere.

[Simon]

Well if you have an actual use for it fair enough. Last time I was asking for advice for my new machine built 7 months ago I had people claiming to be experts and claiming that 64GB on any machine was a must. I am still iusing 16GB like the old machine and can run 3D CAD with lots of browser tabs and circuit studio running as well.

[beanflying]

Did the deal on a 3700X for $312 USD ($491 AUpesos with GST) so just under the MSRP delivered.

After some more looking at GPU's I reached what I think is the correct thoughts is sit on my hands for 2-3 months until Navi and the Super Nvidea triplets sort themselves out on price and stability. To that end I brought a lightly gamed Gigabyte RX580 to tide me over for $167 USD ($239 pesos). Most likely I will look at Card and Monitor upgrades together and resell the RX 580.

So I have failed to meet my $1k budget coming in at $1007

Now all I have to do is stalk the delivery drivers over the next week. Meanwhile cheers All for the input so far, manly Pink Gins all round

[beanflying]

As to be expected a bit of poo flinging at the 3700X against over clocked Intels given the gaming focus of the channel but as a workbeast watch from 14 minutes onward for non game benchmarking of it and some of the other options. 3600, 3900X and some of the I9-xxx's

Not the Odd man Out if you want affordable Workbeast with a bit of gaming on the side and forget about over clocking it is close to max as stock.