• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

AMD THREADRIPPER VS INTEL SKYLAKE X

With the 7900x or above will it be able to handle:

3x3.5 inch sata drvies - My media raid storage
3xSSD - my games and work documents
3xM.2 drives - for VM and encoding work

Maybe put a media encoder card into system, depends how i get on with CPU encoding. I will be gaming, encoding and running vms all at the same time. I do not want to spend 2K on one processor, budget for CPU and motherboard is £1500-£2000.

I will also be creating a lot of how to videos for kubernetes to post to youtube.....

Or will I need to look at ThreadRipper.......

Surely the TR mobo's have the space for 3 M.2 drives? If not then they have the lanes to add a m.2 card and give you the 3 you need.
 
Surely the TR mobo's have the space for 3 M.2 drives? If not then they have the lanes to add a m.2 card and give you the 3 you need.

Most boards do have space for 3 M2s. Zenith comes with extra plug to put 2 of them vertically next to the RAM while the other is hidden near the chipset under a fan (if correct).

Also with TR you can always get PCI-E M2 extension slots and plug 3 of them per slot.
There are plenty of PCI-e lanes to fill them up all 64 :)
 
Not when time is the key. When that's the situation the company simply don't care about power. We supply more binned cpu workstation and servers to these customers than we do binned cpu's to gamers or website customers. We have sone unique oc solutions that tome critical business really likes. Non are AMD and no customer is interested in it because they want fast cores.

This is not to say AMD is not a great option for many it is. They have great products in both ryzen and TR and I will add both to my system range now the memory issues are in the main fixed.
Bit odd that a business would buy a heavily overclocked CPU for a "time critical" function when there is a chance that the overclock could fail. You even said yourself that a small number of chips end up not being stable at their tested settings.
 
Bit odd that a business would buy a heavily overclocked CPU for a "time critical" function when there is a chance that the overclock could fail. You even said yourself that a small number of chips end up not being stable at their tested settings.
I also find odd that said business are happy with how the overclocks are tested. I believe in 8packs video he said 15mins of realbench and 15mins of prime non avx was the testing methods.
If every second mattered to me that much I'd want to make sure it was stable enough to launch a nuke from, therefore I'd get a xeon.
 
Our absolute best silicon goes in these projects as the companys will pay sometimes several times the CPUs retail price. I had one company paying 8-10k for 5960X that would do 5ghz bin and run 4.9ghz for them 24-7.

We dial back from bin by 100Mhz and up the cooling to ensure 24-7 365 stability. Same on the memory. Same on the cache. Our RMA rate on these systems is actually almost zero because the testing is very very thorough in all ways and in line with the cost. Only the best of the best across the board used.
 
I also find odd that said business are happy with how the overclocks are tested. I believe in 8packs video he said 15mins of realbench and 15mins of prime non avx was the testing methods.

Jesus - give it a rest already. We get it, you have an issue with how the systems are tested, no need to jump in at every opportunity to mention it. There's also the concept of "stable enough".


If every second mattered to me that much I'd want to make sure it was stable enough to launch a nuke from, therefore I'd get a xeon.

Xeons don't run at 5ghz though.
 
Jesus - give it a rest already. We get it, you have an issue with how the systems are tested, no need to jump in at every opportunity to mention it. There's also the concept of "stable enough".




Xeons don't run at 5ghz though.

This was before he had explained there are different methods. I assumed everyone had the realbench job. Someone ****** you off today? :)
 
Not when time is the key. When that's the situation the company simply don't care about power. We supply more binned cpu workstation and servers to these customers than we do binned cpu's to gamers or website customers. We have sone unique oc solutions that tome critical business really likes. Non are AMD and no customer is interested in it because they want fast cores.

This is not to say AMD is not a great option for many it is. They have great products in both ryzen and TR and I will add both to my system range now the memory issues are in the main fixed.

When time is key? in the same sentence you are also saying 'Threadripper is ok but none serious is interested in it' so these people are only looking for single threaded performance? is it the 7700K they are buying?
Because if they are looking at something like the core i9 and "time is key" they are looking for the Multithreaded CPU and with that they are wasting time and money with the core i9, the 1950 is the faster Multithreaded CPU no matter what speed the Core i9 is running at.

I know plenty who have seen the core i9 and bought Threadripper because they get the job done faster and save them money on power and cooling costs. "time was key for them" the fact that they also saved money was a bonus.

Dell, HP and others are more than happy to use AMD's CPU's in their flagship workstations, despite AMD in the past taking Dell to court and subsequently fined for taking Intel's $800m back-handers in return for not using AMD's CPU's.

Whatch this review... he benchmarked a very extensive lirary of productivity applications and in just about everything (9 out of 10) the 1950X was faster than the 7900X, in some applications the 1950X was so much faster than the 7900X it looking like Bulldozer vs SkyLake.

 
Last edited:
Why dont they test AVX?

I'm of the view most peeps don't test fully using AVX because it stresses a cpu more than any other test Therefor testing using AVX will almost certainly result in a lower clock. In my case my 1700 is quite happy at 3.9ghz but will only pass OCCT Linpac AVX at 3.8ghz for more than 8 hours. As a result i only run at 3.8ghz on a 24/7 basis.
 
I'm of the view most peeps don't test fully using AVX because it stresses a cpu more than any other test Therefor testing using AVX will almost certainly result in a lower clock. In my case my 1700 is quite happy at 3.9ghz but will only pass OCCT Linpac AVX at 3.8ghz for more than 8 hours. As a result i only run at 3.8ghz on a 24/7 basis.

Which, imo is the sensible way to do it.
If for whatever reason in future a program uses avx (which they are starting to do) and the cpu starts crashing due to an unstable overclock the customer is going to be right on the phone. It would be easier to cover all scenarios imo. Unless of course the end user is using very niche software.
 
I'm of the view most peeps don't test fully using AVX because it stresses a cpu more than any other test Therefor testing using AVX will almost certainly result in a lower clock. In my case my 1700 is quite happy at 3.9ghz but will only pass OCCT Linpac AVX at 3.8ghz for more than 8 hours. As a result i only run at 3.8ghz on a 24/7 basis.

I see so they don't test with AVX because they know the CPUs will fail at the speed they are selling at, so that then means they are knowingly selling unstable overclocked CPUs...
 
I see so they don't test with AVX because they know the CPUs will fail at the speed they are selling at, so that then means they are knowingly selling unstable overclocked CPUs...

Not sure if you're being sarcastic but i highly doubt OCUK are knowingly selling non-stable overclocked CPU's. Apart from the fact it's not a nice thing to do, it makes little sense. RMA's would be a bitch.
 
I see so they don't test with AVX because they know the CPUs will fail at the speed they are selling at, so that then means they are knowingly selling unstable overclocked CPUs...

I didn't actually say what your implying. As 8pack has already said, they dial the clock back 100Mhz anyway. One thing is for sure though, they can't possibly know if any given clock is AVX stable if it isn't tested for AVX stability.
The reality these days is that most peeps on this very forum and other forums, don't bother testing AVX stability either. I put the reason down to laziness and just wanting to get as high a clock as possible for the e-peen effect. Not that many years ago manufactures forums (EVGA and DFI) for instance, had overclocking databases. To be able to register a stable overclock you absolutely had to run a series of tests. One of which was Prime95 Blend test for 24hrs. Now days peeps just can't be bothered to do this, and if they did the clocks that are claimed on here and elsewhere would be much much lower.
 
I didn't actually say what your implying. As 8pack has already said, they dial the clock back 100Mhz anyway. One thing is for sure though, they can't possibly know if any given clock is AVX stable if it isn't tested for AVX stability.
The reality these days is that most peeps on this very forum and other forums, don't bother testing AVX stability either. I put the reason down to laziness and just wanting to get as high a clock as possible for the e-peen effect. Not that many years ago manufactures forums (EVGA and DFI) for instance, had overclocking databases. To be able to register a stable overclock you absolutely had to run a series of tests. One of which was Prime95 Blend test for 24hrs. Now days peeps just can't be bothered to do this, and if they did the clocks that are claimed on here and elsewhere would be much much lower.

Amen
 
Emm,except last time I checked none of the larger clusters in UK universities use overclocked CPUs,including those which were used during the human genome project(at least talking to some people who have worked there a while back).

I mean a minuscule performance for significantly higher power consumption is not what you want,one floor of one of the cluster was basically used for cooling!!

In fact if you look at some of the biggest companies in the world talk about its cooling followed by power consumption. Why??

Because some of these installations are so massive,even the need for extra cooling can add so much power already and some of them consume the power required by a dedicated power plant.

Plus how can you test an overclock properly when these things are on full load 24/7 for months at a time - the downtime itself would be a massive issue.
 
Last edited:
When time is key? in the same sentence you are also saying 'Threadripper is ok but none serious is interested in it' so these people are only looking for single threaded performance? is it the 7700K they are buying?
Because if they are looking at something like the core i9 and "time is key" they are looking for the Multithreaded CPU and with that they are wasting time and money with the core i9, the 1950 is the faster Multithreaded CPU no matter what speed the Core i9 is running at.

I know plenty who have seen the core i9 and bought Threadripper because they get the job done faster and save them money on power and cooling costs. "time was key for them" the fact that they also saved money was a bonus.

Dell, HP and others are more than happy to use AMD's CPU's in their flagship workstations, despite AMD in the past taking Dell to court and subsequently fined for taking Intel's $800m back-handers in return for not using AMD's CPU's.

Whatch this review... he benchmarked a very extensive lirary of productivity applications and in just about everything (9 out of 10) the 1950X was faster than the 7900X, in some applications the 1950X was so much faster than the 7900X it looking like Bulldozer vs SkyLake.

I suspect there's more to it than just the "time is key", ie, more than just the performance of the CPU. At the moment I would guess it's not easy for a company to take a gamble on the unproven Threadripper CPU's. They need proven CPU's here and now, with part availability and low chance of failure and high MTBF. THat's what I'm guessing anyway :).
Besides, people and even companies stick to what they know - it's sometimes hard to break a habit.

It's clear from a few threads that if money is no object, for whatever reasons, still the recommendation is Intel.
 
Back
Top Bottom