How are you planning on setting it up? Using host cache?
Haven't even looked at it yet. That does sound like a good thing to do though. I just thought more faster storage could only be a good thing


Last edited:
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.
How are you planning on setting it up? Using host cache?
It'll only use the drives as write cache for the VM swap files, I'm not aware of a way to target specific VMs.
https://docs.vmware.com/en/VMware-v...UID-4505B56B-57B1-413F-ACE3-BA2A85C3EC88.html
Obviously it would be a bad idea to run production VMs on local storage.![]()
You could purchase some SAS or SATA SSDs along with a vSAN license, I bet that would alleviate your storage issues.![]()
The guy annoys the bejesus out of me, but it was nice to see Linus utterly lose his **** over the 7742.
https://www.youtube.com/watch?v=HuLsrr79-Pw
Apparently he's planning more content based around EPYC Rome too. Given the nature of his audience, I can't help but think it's only a good thing for AMD's mindshare for him to rave so much.
It's an impressive chip for sure. I'm still fighting some bits of infrastructure before being done, it's incredible how much the envelope is getting pushed in the server space with these chips. I'll have a little watch of the video, I normally avoid LTT but will give it a watch.
it's incredible how much the envelope is getting pushed in the server space with these chips
Summary: SuperMicro sent him a board and a 7742. He ran it through Cinebench R15 and R20, BMW Blender render and then ran Crysis using Google's software-only renderer for this teaser/preview. And 256GB RAM for ***** n giggles.It's an impressive chip for sure. I'm still fighting some bits of infrastructure before being done, it's incredible how much the envelope is getting pushed in the server space with these chips. I'll have a little watch of the video, I normally avoid LTT but will give it a watch.
For those of us not clued in, can you unpack this for us please?
Certainly - Think of it this way, Intel can't compete with two of their best in a dual socket configuration vs a single 128 thread 7742. Combine that with 128 pci-e 4 lanes, equal (ish) power draw to a chip that it offers double the performance (plus some) of. Encrypted memory on a per vm basis plus a ton of other features and security built into the silicon.
For hyperscalers and big datacenters literally nothing even gets close when you look at the TCO (total cost of ownership) In the server space this is akin to the kind of performance leap we got going from p4 to core 2 netting big gains across the board while also being a fair chunk cheaper. In terms of density it also finally feels like a proper generational leap.
I'm really loving me epyc cpus so far they offer flexibility I didnt have before at costs we couldn't meet historically. As I was saying proper generational leap in the server space with a ton of extras plugged in for good measure.
Ah, yes.![]()
Personally having almost finished a migration from Intel to AMD I'd be very surprised if Intel can retain the 80% market they want over the next 2 years. They could knock two thirds off of the price and I would bet my last pound that the numbers still wouldn't work in their favour. In the world of virtualisation density and tco are king and Intel are quite a way off.
You would think so, in fact you would think "why on earth would anyone buy Intel now???" but some how no matter how hilarious the comparison Intel always still manage not only to come out on top they keep their domination in market share.
I have no doubt AMD will gain significant market share, and their coffers will do well from it, but Intel will still sell vastly more server chips.
Its partly because of the ease of movement of VMs between processors with a similar base architectures using tech like vmotion (moving a running virtual machine to different hardware with 0 downtime). The problem AMD have is convincing people to change it all or start new clusters. Some people just don't want that aggravation. On a massive scale that could be a pain of a project.
I see, that makes sense.
Well, if AMD sell 50K units that's $400.000.000 In revenue, for Intel that's pocket change, for AMD that's significant, their annual turnover is about <£7BN, compare that to Intel who turnover about <$40BN.
Intel employees 107K
AMD employees 11K
That's what i find really interesting, Intel are turning over about 5X as much revenue as AMD, they employ about 10X as many people.
Can Intel afford to compete with AMD like this?
I think they can, they have a massive war chest so they can probably afford to compete (or not) for many years before anything comes of it. The real question is, can they make something better? Faster, smaller, cheaper and more efficient? And can they do it in a sensible time frame?