Soldato
- Joined
- 19 Feb 2011
- Posts
- 5,849
All of AMDs chips are soldered no? Pretty sure I read Zen was soldered
-
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
Soldato
I'm sure I read AMD are using solder to transfer to the IHS
Associate
Another + for AMD re IHS
Soldato
Not all of them no.
Looking even better for zen then! AMD certainly pushing everything out this time around.
Soldato
All AMD chips are soldered since the transition from the ceramic caps . Not a single one of them used TIM.
Sorry but you're not corect on that.
Apu's since llano used tim and godaveri was the first to.bring back soldered ihs. Also a few am3 processors used tim.
Although I'm pretty sure we haven't had any solid confirmation on if RyZen will be soldered or use TIM. For a long time AMD have soldered the heat spreader, so probably a safe bet soldered zen will be the same
This ^ made me LOL.
I'm not really that interested in gaming performance as my current i7 3930k plays games perfectly fine. As long as Ryzen matches the i7 3930k in gaming then I am absolutely fine. It is the big tasks I care about. Video rendering. Virtual machines. Compiling large C++ projects etc. These are the things that are important to me and hope Ryzen delivers because I really don't fancy spending £4k on a computer when I can get away with spending £3k or maybe even £2.5k.
Associate
Very interested in the 1700x gaming benchies..
Soldato
That's where I am looking too. If it's as good as it looks then it could be a steal with a decent ~£150 board.
Soldato
No it is not 2003.
The OCUK overclocking profile for the i7-920 (or possibly 930) was 191MHz x 23 from my memory to give a 4GHz processor. That showed a degree of experimentation you do not get or need in 2017. I will be very interested in how OCUK and 8 pack approach the Ryzen overclocking and how the community of overclockers here does as well. It will lead to some very interesting discussions I believe.
I did believe that Piledriver had something to offer and it did in terms of a lot of experimentation, but it failed in the main aspect of throughput. Full marks to AMD for trying something different though. I do think that this time they have got it right.
Soldato
- Joined
- 19 Feb 2011
- Posts
- 5,849
All the speculation i have been reading is unfortunately pointing towards Ryzen being not a great overclocker, however worry not because out of the box performance is exceptional. But the general consensus i have read is dont expect huge gains with OC'ing, we wont be seeing any 5ghz+ OC's i believe, i think the 1800X will do about 4.2 on water but the thermals apparently hit hard and fast.
Soldato
Sources for these rumours?
@TaKeN Anadatech
AMD is back, let's not get it twisted, AMD is back and matching INTEL and there's no other way to look at it. Sure you can nit pick here and there, but grand scheme of things, put up an INTEL CPU at 3.4GHz and compare it to 1800X and AMD will win. I can prove this, I have seen this and you all will this coming week.
AMD has extremely good memory bandwidth efficiency, better than INTEL's even in dual channel mode. So I wouldn't say the IMC is weaker per se, yes it's over 1GHZ lower than what INTEL can manage on air cooling (4266MHZ DDR4 vs 3200MHz max on AMD), but weaker is to undersell what AMD has managed here.
So in dual channel mode @ 3200MHZ, CL15 you can get up to 50GB/s memory writes in AIDA 64 which is damn near theoretical limit and that's at CL15.
AMD IMC always runs a command rate of 1T/1N. You can't change it at all and you also can't use really use the tight timings that we can right now on INTEL platforms. Those 12-12-12-26 B-die timings are not going to happen, but no matter, you'll still whip most 7700K systems that aren't tuned to death.
Overclocking this platform is crazy complicated and unrewarding. The clock speeds are low and if you have dreams of a 4.5GHz 24/7 setup, best forget it. In fact most of us will run 3.8 to maybe 4GHz on all cores if we're lucky for 24/7. 1.4V at 4GHz = 90'C load temp with air cooler, so be mindful of that. We need new coolers for the most part because the CPU height is not the same as previous CPUs amongst other changes.
For Haswell-E 5960X to match or beat 1800X (3.6~ 3.7GHzGHz) you'll need about 4.5GHz in the multithreaded programs as Ryzen is ridiculously fast at this beating 6950X sometimes.
And perhaps this follow up
What he means is that a lot of voltage is required for next to no tangible gains. XFR is not guaranteed, it is thermal limited and it needs high end cooling to even push XFR frequency. It is like overclocking a Thuban all over
AMD is back, let's not get it twisted, AMD is back and matching INTEL and there's no other way to look at it. Sure you can nit pick here and there, but grand scheme of things, put up an INTEL CPU at 3.4GHz and compare it to 1800X and AMD will win. I can prove this, I have seen this and you all will this coming week.
AMD has extremely good memory bandwidth efficiency, better than INTEL's even in dual channel mode. So I wouldn't say the IMC is weaker per se, yes it's over 1GHZ lower than what INTEL can manage on air cooling (4266MHZ DDR4 vs 3200MHz max on AMD), but weaker is to undersell what AMD has managed here.
So in dual channel mode @ 3200MHZ, CL15 you can get up to 50GB/s memory writes in AIDA 64 which is damn near theoretical limit and that's at CL15.
AMD IMC always runs a command rate of 1T/1N. You can't change it at all and you also can't use really use the tight timings that we can right now on INTEL platforms. Those 12-12-12-26 B-die timings are not going to happen, but no matter, you'll still whip most 7700K systems that aren't tuned to death.
Overclocking this platform is crazy complicated and unrewarding. The clock speeds are low and if you have dreams of a 4.5GHz 24/7 setup, best forget it. In fact most of us will run 3.8 to maybe 4GHz on all cores if we're lucky for 24/7. 1.4V at 4GHz = 90'C load temp with air cooler, so be mindful of that. We need new coolers for the most part because the CPU height is not the same as previous CPUs amongst other changes.
For Haswell-E 5960X to match or beat 1800X (3.6~ 3.7GHzGHz) you'll need about 4.5GHz in the multithreaded programs as Ryzen is ridiculously fast at this beating 6950X sometimes.
And perhaps this follow up
What he means is that a lot of voltage is required for next to no tangible gains. XFR is not guaranteed, it is thermal limited and it needs high end cooling to even push XFR frequency. It is like overclocking a Thuban all over
Soldato
Heh, memory timings and frequency are intrinsically related. Saying one is good without the other is misadventure. As I recall, there were quite a few people saying memory "bandwidth" wasn't important only last week lol. Broadwell-E is capable of running sub timings close to or event at the minimum spacing required by the chipset. Consistently so, too. I've ran 3 BWE CPU at the same timings with good modules. That's without even beginning to mention SKL/KBL IMC.
Isn't weaker per se, no, it simply is weaker. That said, performance is shaping up nicely - so this isn't going to even remotely stop most. AMD are most certainly back.
Isn't weaker per se, no, it simply is weaker. That said, performance is shaping up nicely - so this isn't going to even remotely stop most. AMD are most certainly back.
Soldato
I am still hoping for some overclockability but if the performance at stock is as good as it sounds then I probably won't want to anyway.