I got a new Asus board recently and it says to populate A2 B2 first.
-
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
I got a new Asus board recently and it says to populate A2 B2 first.
CMU16GX4M2C3200C16
Though the GSkill users have been having better luck, so I'm slightly confused why you're not.
Where did you come across that Naples is two 16C dies? Every news item I've seen which has mentioned anything has said it's four 4C dies in one package. Like for instance:
http://www.anandtech.com/show/11183...aples-cpus-for-1p-and-2p-servers-coming-in-q2
Which makes sense as testing and making masks is getting expensive and also in terms of yields etc. AMD are still a very small company compared to Intel and this is a far cheaper way to make big dies. This plus the fact that Ryzen is a SOC should all mean that making a 16C/32T chip shouldn't cost them too much. Two times Ryzen should even have enough IO to not need a chipset. The Naples dual-CPU example motherboard is very bare:
Well, more or less. But hardware.fr found differences between the voltages needed for a given frequency (as did SL):
as did the BitAndChips.it in their underclocking tests:
So, yes the Ryzen 5 1600 sounds like the best value, but clocking it at the speed of the 1600X will possibly cost a few more watts. But then in theory it should be a lot lower than 65W at stock being a 'smaller' chip. (What exactly happens to fused of parts in terms of power draw?)
Indeed, I can only conclude the IMC is weak, sadly.
I've got a set of that RAM but the 32Gb version, it's the LED one, right?
It's unopened and the RMA has been approved so I can't really open it to try.
Maybe you don't care about watts, but do you care about managing the extra thermal that those watts are giving you thanks to the higher voltage?
Yeah but the literal watts being discussed are so small we are talking a couple of degrees not even 10 or 15 degrees.
Indeed. My point wasn't really about the watts so much the expectation that 1600 and 1600X will clock the same as there clearly are binning differences between 1700 and 1700X so I would expect something similar. Perhaps even worse if they are die-harvesting the hex and quad cores from actual failed octo cores ones. Or at least with the expectation that some of the R5 parts might require higher voltages which would make the octo cores jump past their TPU.
Like I said, I don't know who fused-off parts act and Ryzen is a SOC, but a quad core part should use a lot less than a octo. If you take that 4.0GHz 1800X @ 1.35V from that Italian site and subtract 40W for the idle platform, you are left with 172W and assuming a 90% efficiency from your PSU you get down to about 155W. So ignoring VRM losses, and assuming that the SOC part draws 10W (although I suspect most of that is already in the 40W platform idle I took off), you have about 145W and half of that (the hypothetical fused-off quad part) gives around 78W. Still way under the TPU they are quoting for R5 despite 4.0GHz @ 1.35V using an additional 40W compared to stock.
Point being for the hex and quads they could afford to pump higher voltages and still stick to their TPU. There is a reason it's called the silicon lottery.
with my external radiator even when im pumping constant 1.45 in to ryzen chip it maxes after hours @72c so
Cant beat overkill cooling that i built for cooling down 2500k@5ghz running constant 1.49 vcore that was putting out 200% of heat that overclocked 1700x does 
If i remember it was drawing around 260watts(CPU Alone !!!!) on heavy benchmarks
Last edited:
likely not till the 11th if they do the same as with the ryzen 7s
what You said. 4.1 max
Caporegime
- Joined
- 18 Sep 2009
- Posts
- 30,294
- Location
- Dormanstown.
We don't really know what they'll clock to.
You'd hope with a lower core count that it'll be even 200MHZ higher on average.
You'd hope with a lower core count that it'll be even 200MHZ higher on average.
Deleted member 66701
Deleted member 66701
Disabling cores on the existing R7's shows no improvement though, so I doubt it.
That would be my guess to. What's holding it back is GF's/Samsung process not the design. Unless they port it to another process I doubt we'll see much above 4GHz and the voltages required for that are already way outside the process's sweet spot. Not so good for IPC clock junkies although for server and mobile the present process is very good indeed.
The questions then are:
- does AMD have access to faster clocking (i.e. not designed for mobiles) process
- would it worth AMD's while to port Ryzen to such a process considering that most of the volume market should happy at under 4GHz