Not if it's 50% less performance2080Ti to Vega64, £1200-£450 so 37% of the cost of a 2010Ti the Vega64 looks a good deal now doesn't it
-
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.
Core 9000 series
- Thread starter orbitalwalsh
- Start date
I would agree on principle sure.. Manufacturing gets better all the time, yealds improve as they iron out more issues from the process. Intels 14nm+++ with the 9900k is about topped out, they did very well getting an 8 core CPU to 4.7GHz on 14nm, but you can tell that it's very near the limit just by the amount of extra heat it'll put out just by increasing the clock speed by 100MHz. Getting it to 5GHz, so +300MHz from the 4.7GHz all core turbo, is a massive increase in the heat output. If you can get one stable @5GHz you'll need it to be under water, not sure of the output but at a guess I'd say 250-300W.
I think, hope, Zen2 will kick intels' arse, and I don't think they can have a reply to it for 6 months at least.
Caporegime
More given that they are not going to get 5Ghz+ on 10nm for a long time. i think its going to be years before Intel catchup with themselves on 10nm.
Or just over-promising marketing.
Soldato
Looking at in game performance I would take the 2700x and pocket the £300 saved, less of a fire hazard as well.
Caporegime
It is true GF 14nm was a 3Ghz process, it simply means its where the silicon is most efficient, like the 8 core 16 thread 1700 at 65 Watts, that's the 3Ghz chip.
You can of course go higher but it costs power, the higher you go the better quality silicon you need to keep reasonable power levels.
GF 12nm promised 10% higher clocks at the same power levels, that's just about what we got with Ryzen 2000.
GF 14 and 12nm are designed for very low power chips, mobile chips, not for high clocks.
Now to get onto B12B6 point, it was not TSMC who said 7nm would be 5Ghz silicon, it was GF, they are no longer doing 7nm, AMD have moved production to TSMC and we don't know the speed ratings of their silicon, i think its safe to say tho that TSMC 7nm are rated higher than 3Ghz and 3.3Ghz.
You can of course go higher but it costs power, the higher you go the better quality silicon you need to keep reasonable power levels.
GF 12nm promised 10% higher clocks at the same power levels, that's just about what we got with Ryzen 2000.
GF 14 and 12nm are designed for very low power chips, mobile chips, not for high clocks.
Now to get onto B12B6 point, it was not TSMC who said 7nm would be 5Ghz silicon, it was GF, they are no longer doing 7nm, AMD have moved production to TSMC and we don't know the speed ratings of their silicon, i think its safe to say tho that TSMC 7nm are rated higher than 3Ghz and 3.3Ghz.
That is a very good point. Intel is now on the fourth iteration of the 14nm process, 14nm, 14+, 14++,14+++, yields and clock speeds have increased way beyond the original 14nm projections. So when Intel finally starts 10nm production, yes I know intel did start 10nm production on a laptop part with disabled graphics but canned it soon after, they may actually run at a lower clock speed then previous generation chips. And I think that is a real possibility!
I've read that Intel have reduced the specs on 10nm just to get it out of the door producing chips so that also possibly won't help with speeds either. If you think about it, if Intel had converted over several fabs to 10nm for the process not to have been approved you have all of that fabrication capability just sitting there idle. I'm not sure if you can use a 10nm fab to produce 14nm parts but I wouldn't have thought so, so Intel are possibly sitting on billions tied up in fabs that they cannot use even for 'old' parts. And now they don't have enough 14nm supply, I bet that some massive internal milestones were missed, including contingency, so we have Intel in the mess they are at the moment.
Soldato
- Joined
- 31 May 2009
- Posts
- 21,262
You'd think they'd sack their CEO...
Oh yeah...
Oh yeah...
Caporegime
Again its never more than a 15% difference, Nor does Intel always win.
https://youtu.be/n8QRaYGq4dk?t=260
It would be costly and involve a lot of time (potentially months and months) doing validation to produce say 14nm at a 10nm fab - when Sony did something like that it took 13 months before they could start production and then probably the same again to revert to 10nm production.
Associate
Again its never more than a 15% difference, Nor does Intel always win.
https://youtu.be/n8QRaYGq4dk?t=260
I watched that review last night, and it was the only review that seemed in multiple games to have ryzen consistenly doing as well or in some cases better than the i9 - most other reviews seem to have it 20% behind in performance. So I'm not sure this one is all that reliable.
Soldato
Look RAM is the main factor here. The significant majority of reviewers used some pretty pathetic RAM (2993/3000).
Joker even if he didn't activated DOP/XMP used really good ram with 3200C14. Which gives a hefty 10% perf to stock 2700X over something like 2933.
Same guy has Vega 64 v 1080 video of supposedly "stock" cards but the Vega 64 is holding clocks it won't do "stock" though IMO fair enough as you can attain them with a slight tweak and no real overclocking involved and if you have a good core can even drop the voltage slightly for a big power saving. But his 1080 clocks are down 50MHz on a typical stock 1080 and if you applied the same level of "tweaks" as the Vega card obviously has you'd get around 100MHz higher clocks.
So yeah...
EDIT: I'm actually all for showing products in their best light though it can be subjective where you draw the line but it needs to be applied equally.
Last edited:
Caporegime
All of them on bar chart slides. even then not all of them, some are more like 10%
You see i would argue the only ones you can trust are the ones where you can see the two playing out side by side, anyone can make a slide bar up and put anything they like on it, having the game actually running before your eyes with MSI OSD is very difficult to cheat.
Tho he has not yet done the 9900K... Here is another one who does this and surprise surprise he gets the same sort of results as JP 8700K vs 2700X.
If these guys can do it why do the rest still insist on sticking with bar chart slides only?
Caporegime
All of them on bar chart slides. even then not all of them, some are more like 10%
You see i would argue the only ones you can trust are the ones where you can see the two playing out side by side, anyone can make a slide bar up and put anything they like on it, having the game actually running before your eyes with MSI OSD is very difficult to cheat.
Tho he has not yet done the 9900K... Here is another one who does this and surprise surprise he gets the same sort of results as JP 8700K vs 2700X.
If these guys can do it why do the rest still insist on sticking with bar chart slides only?
Another example of this.
https://youtu.be/4RMbYe4X2LI?t=318
Saw this comment on youtube that cracked me up
14nm+++++++++ is why it is so hot. They need to get on 10nm asap.
Funny thing is, probably because they assumed they'd be moving in to 10nm before it was worth it, they skipped some optimisations on 14nm despite being on a "+++++" variant. They can probably get another 15-17% (I'd have to look it up for the exact figure) from it yet - I don't think they've done it for the latest CPUs.
look at numerous different benchmarks. dont use one. many people are pro amd or pro intel. the amd cpus are about 20 percent on avg behind the new intel i9s in games. crysis 3 for eg 9900k vs 2700x = 35 fps difference. games that use the power you will see big differences.
Soldato
When the 2700X is crippled with sub par RAM. Only Joker used good RAM and suddenly his benchmarks are not good enough.....