Sounds to me like extrapolation and speculation based on the 16c/32t Naples chip.
-
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 "Zen" thread (inc AM4/APU discussion) ***
- Thread starter Boomstick777
- Start date
The same people that buy X99 will buy them. X99 is the still the best workstation platform as it has more RAM slots and PCI-E lanes and can also take high core count Xeons. AMD should have a competing platform.
Soldato
- Joined
- 18 May 2010
- Posts
- 23,139
- Location
- London
I don't think it takes AMD much effort to slap two 8/16 thread CPU's together using their infinity fabric.
Meh, I moved to Ryzen because the entire setup was cheaper than a 6900K.
I'm not saying there isn't a market, I'm saying the market is so microscopic that if this rumor is true they will be lucky to make enough to cover R&D.
Interesting he said the high RAM speed does change IPC, just leverages the power of the chip, as if slower RAM is holding it back.
I think the market is bigger than you think, Intel are doing ok with X99. I could not use my Ryzen PC for work as it does not have enough PCI-E lanes. I need lots of fast storage as it speeds up testing a lot. The AM4 platform is AMD mainstream, they need to offer a workstation option.
Most likely due to Infinity Fabric is linked to memory speeds: https://www.reddit.com/r/Amd/comments/5zr8lv/i_asked_amd_a_followup_question_about_infinity/
So faster memory = faster IF = faster movement between CCXs. Which pretty much "solves" the gaming issue with Ryzen.
the size of the market is dictated by the entry price, cut intel's cost by half and you suddenly multiply the market size by x10.
intel 10 cores 1700$ + x99 500$ <--- this is hardly worth considering, unless you absolutly need it.
AMD 16 core 900$ + x399 200$ <--- this is worth considering, even if you dont really need it
it is actually better for them to use one of the 16c, 64PciE, 4 mem controller dies they use to make 32 thread naples.
32c naples is made of two 16c Dies in an MCM package.
I can't view the video till later.. so making an assumption here: Didn't some reviews already test 3200Mhz on RAM and it didn't solve the gaming results? What memory speeds are shown in this video to solve the lagging gaming performance?
That's obvious. You seem to think a chip has 'an IPC', but every operation takes a different number of clock cycles depending on many factors, including the memory access (to CPU cache and/or system memory). Doing a lot of reading/writing to RAM? Then RAM speeds will dictate IPC.
It is indeed and has been previously discussed. It means that as the bios updates to allow faster memory which will help with non-gaming situations that the Infinity Fabric limitation will also be resolved for gaming.
What I don't understand is why they didn't put a dedicated cache of memory to support the infinity fabric speeds and thus offer the 512GBps that they will have with the Volta GPU's. That could also significantly increase the speed and remove dependency on the system RAM.
Permabanned
- Joined
- 12 Sep 2013
- Posts
- 9,221
- Location
- Knowhere
It sounds good but not of much help when you have to artificially hold the i7 back to match it,
Unless Ryzen gets better at core overclocking in future iterations it's going to run slower than Intels chips regardless of what faster ram achieves with it.
Soldato
- Joined
- 18 May 2010
- Posts
- 23,139
- Location
- London
Does it? But then you've got the issue of DDR4 compatibility issues with Ryzen.
If the mobo is rated up to 2400 and you buy 3200Mhz ram, will it run at that speed or will you have to manually overclock it?
Yeah but what it is stating is that the IPC is actually good. Just need Ryzen to mature with time and that was always expected. No one I don't think has still suggested that Ryzen will be on par with the 7700K where it isn't GPU bound and you are playing a game that utilises no more than 4 Cores.
That though becomes much more difficult to show what can be done and where it is at otherwise. What it should show is where like in the BF1 multiplayer where the 6900K beat out Ryzen at the same or very similar clocks is that Ryzen should be on par with that also because the loss in the system for Ryzen is currently the Infinity Fabric and RAM speed issue which the Intel chips do not have.
That means where you see the advantage in the 6900K in gaming (although not often will improve over time) compared to the 7700K or similar then you will also see that Advantage much more obviously in a Ryzen setup. Add that to the fact that it should be much more on par with the scores the Intel enthusiast chips are getting then it shows that it does make a tangible difference that is important for the future going forward.
1.14 at the moment. I won't update until one is released with extra features.
Manually overclock/apply XMP profile, hence why I placed the "solves" bit in quotation marks, because it's still a mess getting around the issue and we still need that BIOS update for the high memory speed fix. But still strange AMD locked the IF speed to 1:2 of the memory speed.