AMD's Future Chips & SoC's: News, Info & Rumours.

[boju]

Charge less sell more?

 

[InvalidError]

Charge less sell more?

Unsold chips don't generate revenue and if Intel does not do something about stale inventory, then it may end up with the even more costly prospect of having to write off some inventory altogether on top of lost sales.

Intel hasn't had to worry about viable competition for over a decade, Ryzen is bound to force some adjustments in how Intel does business.

 

[NightHawkRMX]

Ryzen has changed a lot of things for the better.

 

[boju]

Unsold chips don't generate revenue and if Intel does not do something about stale inventory, then it may end up with the even more costly prospect of having to write off some inventory altogether on top of lost sales.

Intel hasn't had to worry about viable competition for over a decade, Ryzen is bound to force some adjustments in how Intel does business.

I meant prior to overstock, why couldn't intel sell for less? Share holders? Stubbornness? Fat heads? Intel deserve their right wack if anything of it relates. Even if superior, doesn't give anyone the right to over inflate prices, just like petroleum.

 

[InvalidError]

I meant prior to overstock, why couldn't intel sell for less?

Simple capitalism: if your mass-produced products sell to the point that you struggle to keep up with demand, then your prices are too low.

 

[boju]

Simple capitalism: if your mass-produced products sell to the point that you struggle to keep up with demand, then your prices are too low.

Make customers wait coz it's worth it ;P Na i know its more complicated than that and price of business, but ya know, a little less expensive wouldn't hurt anyone really.

I appreciate your feedback InvalidError

 

[Deleted member 14196]

I can’t wait for the 16 core cpus are released because we are going to do a mass upgrade at work and shrink the amount of VMware workstations we have set up to just a few boxes

We don’t have the infrastructure to run the big hypervisor’s so we use individual workstations to host about 10 VMs per workstation but this way we could even fit more VMs on one workstation with tons of RAM

We are figuring we can probably reduce our VMware workstations to less than half of what we currently run so I am excited like Borat

 

[NightHawkRMX]

I can’t wait for the 16 core cpus are released because we are going to do a mass upgrade at work and shrink the amount of VMware workstations we have set up to just a few boxes

I hope the memory badwidth doesnt hurt the high core count cpus too severely.

 

[Deleted member 14196]

I hope the memory badwidth doesnt hurt the high core count cpus too severely.

Me too buddy. I feel like Borat when he was interviewed on television and he’s so excited he can’t even talk

 

[InvalidError]

I hope the memory badwidth doesnt hurt the high core count cpus too severely.

If getting the most performance out of 16 cores in memory-intensive tasks is a major concern to you, then ThreadRipper 3 might be a better option.

 

[Deleted member 14196]

If getting the most performance out of 16 cores in memory-intensive tasks is a major concern to you, then ThreadRipper 3 might be a better option.

I will present this. If it saves us money vs the new Ryzen then it will be a great thing

Thanks guys

 

[NightHawkRMX]

For juat a few virtual machines, is ram bandwith really that important?

 

[Deleted member 14196]

The software we write and test is very memory intensive and many of these VM’s are going to be testing our products lots of network activity and lots of memory access.

It would be nice to be able to run say at least 50 VM’s on each box with nobody complaining of performance issues

Some are just development VM’s and I have a couple build VMs out there

 

[NightHawkRMX]

I just use vms lightly. I can run a few vms simultaneously on my current quad core, so a 16 core would certainly be able to run a ton.

 

[InvalidError]

For juat a few virtual machines, is ram bandwith really that important?

It depends entirely on what is running in the VMs. Memory-instensive software is memory-intensive regardless of whether it runs on the host OS or VM.

 

[Deleted member 14196]

I just use vms lightly. I can run a few vms simultaneously on my current quad core, so a 16 core would certainly be able to run a ton.

A typical test VM is two cores and two gigs of RAM Running windows 10 or any modern Windows operating system.

However one core would be sufficient for a test machine. I think we would need epyc servers to do what I’m thinking. That being said the thread ripper is looking better and better and better

 

[NightHawkRMX]

I read somewhere that Cinebench relies heavily on the floating point.
So is that why AMD keeps showing Cinebench scores even when they don't have a core count advantage. Since they "doubled the floating point" with 3rd gen ryzen?

One thing surprised me is how well 3rd gen ryzen CPUS do in cinebench R20 as well since R20 uses AVX, something current gen ryzen kinda sucks at.

 

[jdwii]

I read somewhere that Cinebench relies heavily on the floating point.
So is that why AMD keeps showing Cinebench scores even when they don't have a core count advantage. Since they "doubled the floating point" with 3rd gen ryzen?

One thing surprised me is how well 3rd gen ryzen CPUS do in cinebench R20 as well since R20 uses AVX, something current gen ryzen kinda sucks at.

Dolphin emulator uses this instruction set its the main reason Haswell was so much faster in Dolphin vs sandy and ivy.

 

[NightHawkRMX]

Ryzen does support AVX2 just like Haswell onwards, however, its performance isn't great while using that instruction.
Ivy bridge and sandy bridge do not support AVX2, but they support AVX. Nelhelm CPUs dont support any form of AVX

 

[aldaia]

Not in the mainstream. At the moment, the Cortex A65E (automotive) is the first and only SMT ARM core design. Give ARM a few more years to run into all the same IPC brick walls as everyone else, then they'll go SMT across most of the product stack as a more power-efficient way of getting more work done per clock per core.

Was not aware of Cortex A65.
Besides that, the Broadcom Vulcan was SMT4 https://en.wikichip.org/wiki/cavium/microarchitectures/vulcan
The Vulcan design was later acquired by Cavium. The thunderX2, based on the vulcan core, is also SMT4. https://en.wikichip.org/wiki/cavium/thunderx2

 

[jaymc]

A few things about the test...
https://pbs.twimg.com/media/D9bJldjXkAE-Nqw.png

1. PassMark takes about ~1 min. 30 seconds to complete give or take, so it's possible the i7 could boost at 3.9GHz for the entire test, 200MHz higher than the boost of the 3500U.
2. Why compare an i7 to the R5 3500U instead of the R7 3700U?
3. A comment was made on twitter that the source of the test was fake, I'm checking on that now.

Edit: 3. "Chiphell user posted it then it was deleted eventually" So, likely a fake.

Well after reading that it sounds fake or a feeble attempt at regaining some mind share that backfired due to obvious bad comparison's you mention.
Agree with jdwii though an IPC war would be nice, Now that the core wars are over that is 😷
Still seam's like there really struggling with Ice Lake, any decent silicon will or would hit the data centers first.

 

[jaymc]

If AMD is starting to eat into OEM orders and Intel has sorted out its shortage issues, then Intel may be seeing some surplus inventory. There are no miracles to mitigating surplus stock, you either reduce production or lower prices to stimulate demand and make the surplus go away. Extra manufacturing capacity does you no good if the extra chips only end up hogging warehouse space.

That's what I was thinking, they are sinking billions into their FAB's, 8b in Ireland alone to Increase production capacity. By the time there finished the demand may no longer be there. Not just due to ROME, but also company's had to upgrade just to compensate for loss of productivity due to Intel Exploit Fixes. Another reason for the excessive demand all of a sudden, if true also lends weight to a temp Increase in demand, on both counts from Intel's pov anyway. Their firefighting an appear to be a few steps behind right now.

 

[jaymc]

Unsold chips don't generate revenue and if Intel does not do something about stale inventory, then it may end up with the even more costly prospect of having to write off some inventory altogether on top of lost sales.

Intel hasn't had to worry about viable competition for over a decade, Ryzen is bound to force some adjustments in how Intel does business.

He may be referring to AMD's ability to charge less for more. But what volume can they meet. If adoption were to suddenly swing towards AMD, how long would it take to increase or ramp... they claimed to be aiming for 10% of Server market (actually watered down deadline now) to early 2020 instead of EOY-19. So adoption already looking slower than AMD expected previously.

 

[goldstone77]

Interesting sneak peak at Ryzen 3600.
https://pbs.twimg.com/media/D92cCHSX4AETH_z.jpg
https://pbs.twimg.com/media/D92cDn6WkAAelsQ.jpg

Review: AMD Ryzen 5 3600 (X470) (Exclusiva Mundial)

Como viene siendo costumbre con las novedades de AMD, nos adelantamos a su lanzamiento oficial para traeros una jugosa review del nuevo AMD Ryzen 5 3600.

elchapuzasinformatico.com

 

[goldstone77]

Another 3600 that popped up, using 4233MHz RAM
https://www.userbenchmark.com/UserRun/17880340