Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
AMD Discussion
- Thread starter koyoot
- Start date
- Sort by reaction score
I know they're not the most trustworthy site out there but still I believe the lineup makes sense: 8c/16t, 8c/8t and 4c/8t. No room for 6c.
Its gotta suck to be a motherboard manufacturer. In the enthusiast space you are only competing on who has the shiniest LEDs.
Performance-wise I prefer Nvidia and that's what I own.
But the overall market is a different beast, and AMD has been clawing back market share in the low and mid end range. The result is that AMD stock made me a ton of money. Sold it in December, well after Polaris was released and well-known.
There is good news, however. It appears that the best motherboards for Ryzen platform will cost between 150 and 200$.
This is biggest clue for the end price of Ryzen CPUs we have had till this day
.P.S. One of the rumors floating around, after todays presentation: AMD truly wants to offer 4K gaming setup for under 1000$ price point.
I would imagine that by now the LED fetiche would have been gone but hey... boys will be boys!!
If there will be one in Mini-ITX form factor: sign me in
.NZXT has pretty cool looking fans with LEDs included
One more thing:
AMD said:
Last edited:
Well, not gonna say I told you so
https://www.techpowerup.com/230227/amd-zen-cpu-complexes-indivisible-dont-expect-6-core-ryzen-report
https://www.techpowerup.com/230227/amd-zen-cpu-complexes-indivisible-dont-expect-6-core-ryzen-report
I don't care about LEDs.
I care about things like:
- 10GbE
- 2x M.2 22110 slots
- Thunderbolt
- Long life capacitors
- Best RAM clock support
- Layout
- Number of SATA (Express) ports
- "Optimal" expansion slot configuration
Techpowerup made a note from "reports"(rumors) floating around.
If information from french Magazine(CPC) has any truth in it, and they have had/seen Engineering samples, there is 6C/12T CPU, already going through validation, and it has 65W TDP.
If information from french Magazine(CPC) has any truth in it, and they have had/seen Engineering samples, there is 6C/12T CPU, already going through validation, and it has 65W TDP.
Last edited:
https://twitter.com/CPCHardware/status/826829044402552833
AMD valide bien actuellement des Ryzen 6C/12T @ 3.3 GHz base mais PAS de 8C/8T. Le but est clairement de recycler les dies 8C défaillants.
In essence for non French-language speakers: AMD is validating the 6C/12T CPU, but not the 8C/8T ones. Its all about getting as high number of failed 8C dies, as possible.
AMD valide bien actuellement des Ryzen 6C/12T @ 3.3 GHz base mais PAS de 8C/8T. Le but est clairement de recycler les dies 8C défaillants.
In essence for non French-language speakers: AMD is validating the 6C/12T CPU, but not the 8C/8T ones. Its all about getting as high number of failed 8C dies, as possible.
Last edited:
Skylake or kaby lake does not have the pci-e for that!
amd zen maybe 10GBE may have to share with m.2 slot 2
About stock price, and my previous comment:
It will be interesting to see in upcoming months what will happen. And remember. In January 2016, AMD stock was 1.84$. 1.84!
https://investorplace.com/2017/01/advanced-micro-devices-inc-amd-stock-turnaround/#.WJITdT197hMSpeaking of stock. You have to be soft in the head to not buy the AMD stock right now
It will be interesting to see in upcoming months what will happen. And remember. In January 2016, AMD stock was 1.84$. 1.84!
koyoot, maybe they can still use the good cores in a damaged package.
Joe, depends on what you are considering. Desktop parts, -X or -W? The CPU might not have enough lanes but the PCH will handle it. But then you have the DMI bottleneck. But it's possible, might not be perfect but possible it is.
Joe, depends on what you are considering. Desktop parts, -X or -W? The CPU might not have enough lanes but the PCH will handle it. But then you have the DMI bottleneck. But it's possible, might not be perfect but possible it is.
CPCHardware claims that the only CPU with the same amount of threads as cores, at the beginning is the 4C/4T setup.
Manuel, I am not saying that anyone is right or wrong. I just look at the information from the people who actually have had any form of contact with the actual Ryzen Silicon.
Zeppelin design is actually pretty modular design and it can scale from 2C/2T to 8C/16T on single die, without any problem.
Manuel, I am not saying that anyone is right or wrong. I just look at the information from the people who actually have had any form of contact with the actual Ryzen Silicon.
Zeppelin design is actually pretty modular design and it can scale from 2C/2T to 8C/16T on single die, without any problem.
Also, there is only ONE design. So if you will see 2C/4T, be sure that there are 8 CPU cores, and 6 of them had failed validation. Each die costs AMD precisely 18$ at this point.
AMD wants to sell as much dies as possible. They will charge more than 450$ for top end SKU, however, IMO maximum price we will see is 500$. But that is just, and only, my personal opinion.
It appears also, that not in every situation Ryzen will be as good as hyped by AMD. Average IPC is Broadwell level. In compute scenarios without AVX instruction - its actually faster. In games, it will be EXTREMELY competitive with the best Intel offers. However there are some cases where the IPC is on Ivy Bridge levels.
AMD wants to sell as much dies as possible. They will charge more than 450$ for top end SKU, however, IMO maximum price we will see is 500$. But that is just, and only, my personal opinion.
It appears also, that not in every situation Ryzen will be as good as hyped by AMD. Average IPC is Broadwell level. In compute scenarios without AVX instruction - its actually faster. In games, it will be EXTREMELY competitive with the best Intel offers. However there are some cases where the IPC is on Ivy Bridge levels.
That's not what asymmetric multiprocessing is.
And in an SMT processor, the vast majority of the transistors are shared between the two threads - just the register files and some other state are unique to the thread. It's unlikely that there would be very many chips where all of the shared transistors are fine, but there are errors in the small amount of per-thread state.