This for me also. Although I won't be buying <8 cores
-
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 2 (Ryzen 3000) - *** NO COMPETITOR HINTING ***
- Thread starter Jamin280672
- Start date
Soldato
- Joined
- 22 Nov 2018
- Posts
- 2,840
I'll be happy with whatever CPU I can get for around £175 ish. Hopefully the 3600 but it's release price will probably be £199
Last edited:
Permabanned
- Joined
- 12 Sep 2013
- Posts
- 9,221
- Location
- Knowhere
If those cores & numbers are true (which I seriously doubt) & if it's anything like the 2700x we'll be looking at 5.0 at best.
Permabanned
- Joined
- 12 Sep 2013
- Posts
- 9,221
- Location
- Knowhere
We know from Lisa Su that there's at least one cpu in the line-up with more than 8 cores so if pricing remains sensible I'll be looking at that and a new motherboard to replace my X470 m/b & 2700x to put in a new build this summer.
If those cores & numbers are true (which I seriously doubt) & if it's anything like the 2700x we'll be looking at 5.0 at best.
Thought the leaked specs were 5.1Ghz boost for the 3850X?
Soldato
- Joined
- 22 Nov 2018
- Posts
- 2,840
Lisa Su only had an engineering sample at CES and those specs leaked out months before that. There's no way they knew what the exact final frequencies would be back then. They'd be an approximation at best.
Given the amount of time, I am inclined to agree. Let's just cross our fingers these chips have legs, and it isn't another Polaris "power" demo.
It goes without saying that you're welcome to make your own judgment on how accurate the leak is but speaking personally i too find them hard to believe, an extra 800Mhz would be quiet a leap especially when so far products based on the 7nm node have typically seen increases in the 200Mhz range, that not to say we'll only get a 200Mhz bump as what's been released on 7nm so far have (afaik) been a GPU and an ARM based SoC so a CPU could well have higher thermal thresholds.
Personally I'm expecting something more around a maximum in the 4.8/9Mhz range.
Associate
- Joined
- 7 Apr 2017
- Posts
- 1,762
4.8ghz would still likely take the single core performance crown with the IPC gains, but no reason to think these chips wont boost well, as it's not on an adapted mobile node this time around. I'm expecting good clocks and all core overclocking potential.
I do wish people would stop conflating IPC with performance and clock speeds, instructions per cycle is a fixed number that only changes when you make architectural changes and more importantly it only applies to very specific workloads, in other words it's largely meaningless when it comes to comparing real world usage and even more meaningless when comparing CPU a with CPU b.
Putting my little moan aside when i said i suspect 4.8/9Ghz i was talking about maximum clock speed i.e boost clock under non-exotic cooling, that's still a good speed for a single core to be boosting too IMO as clock-for-clock Ryzen typically performs better than Intel's Core design (I would say it has a higher IPC but it doesn't because IPC measures a very specific workload and they're near identical when measuring that), yes Intel's core design is faster on a clock-to-clock basis when it comes to certain code/workloads but overall Zen performs better in more situations.
Putting my little moan aside when i said i suspect 4.8/9Ghz i was talking about maximum clock speed i.e boost clock under non-exotic cooling, that's still a good speed for a single core to be boosting too IMO as clock-for-clock Ryzen typically performs better than Intel's Core design (I would say it has a higher IPC but it doesn't because IPC measures a very specific workload and they're near identical when measuring that), yes Intel's core design is faster on a clock-to-clock basis when it comes to certain code/workloads but overall Zen performs better in more situations.
Permabanned
- Joined
- 12 Sep 2013
- Posts
- 9,221
- Location
- Knowhere
While I don't hold stock in that type of leaks credibility if it was like the 2700x we'd see lower clocks than stated not higher, The 2700x was officially quoted as boosting to either 4.3 or 4.35 on a single core, I forget which, but for overclocking 4.25 to 4.275 is the best I've seen shown as stable, Mines at 4.25 under an AIO and it has good temps, I can boot it up with a 4.3 overclock but once it's running at load it crashes, I haven't seen better with proof anywhere. Plus we see unreliable results as these show where my overclocked to 4.25ghz cpu is reported as running at a 4.3, 4.4 & even 4.7ghz. https://www.3dmark.com/compare/fs/18261261/fs/18352449/fs/18352509# These are all unreliable results not true examples of Ryzen's potential so rather than be disappointed we'd all be better off keeping our expectations in check. I'll be very happy if I can get a cpu capable of an all core overclock at around 4.7 or 4.8, and that's where I'm expecting the flagship to be.
Soldato
- Joined
- 22 Nov 2018
- Posts
- 2,840
Its possible that 4.8GHz is all that's needed to beat a 5GHz Intel CPU anyway. Maybe even 4.7GHz could beat Intel if the IPC is good enough.
It's just the human brain's need to have this big round number to make us feel better.
It's just the human brain's need to have this big round number to make us feel better.
I firmly believe AMD will be busting a gut to release a chip capable of 5ghz + boost - they need to make further inroads into Intel's mind/market share and even if the IPC is better and the chips perform better than the 9900k the great unwashed will still look at the numbers and think "intel=better" so in my opinion it is imperative for AMD they hit above 5ghz and I personally think this will happen. The node is a high performance node and there has been information released from TSMC quoting 5ghz performance reports on their 7nm.
Well my 8600k hits 4.8ghz stable at sane voltages, the feeling that 5ghz is an easy o/c in my view is misleading. I expect many may have their chips at 5ghz but with implications such as using avx offsets (cheating) or perhaps considering been able to boot into windows and do some light gaming as to considering the clock stable, whilst not passing any proper stress testing.
I would possibly consider replacing my 8600k with AMD if they can hit 4.6+ clocks. I dont think they would need to hit 5ghz. As I think their IPC when taking into account meltdown performance hits will be better. I think the performance impact of meltdown got massively understated by the media. You can see reports on reddit when people disable meltdown they fixing things like stutters in games, getting in some games significant performance improvements as well as better responsiveness on their systems, my experience kinda mirrors that. When I enable meltdown mitigation on my rig, it feels like I am cancelling the haswell to coffee lake upgrade I did, the impact is that big.
On my i5 750 esxi rig when I had meltdown mitigation enabled it was adding a 2-3 sec pause to simply loading the start menu in a windows 10 guest, disabling the mitigation took it down to about 1 second, so the overheads were clearly large although thats an older chip that doesnt have PCID enhancements (its instant like bare metal on my 2600X). When people talk about it only impacting i/o they assume if you not doing i/o to a storage device then its no impact, but ram access is also i/o. Cpu's are constantly doing i/o to ram.
For hypervisor workloads the performance impact on intel chips is huge, if you enable every possible mitigation you can wave bye bye to circa 50% of your cpu grunt on an intel platform, I am not joking, in that case a 2.5ghz AMD chip could match a 5ghz intel chip.
I would possibly consider replacing my 8600k with AMD if they can hit 4.6+ clocks. I dont think they would need to hit 5ghz. As I think their IPC when taking into account meltdown performance hits will be better. I think the performance impact of meltdown got massively understated by the media. You can see reports on reddit when people disable meltdown they fixing things like stutters in games, getting in some games significant performance improvements as well as better responsiveness on their systems, my experience kinda mirrors that. When I enable meltdown mitigation on my rig, it feels like I am cancelling the haswell to coffee lake upgrade I did, the impact is that big.
On my i5 750 esxi rig when I had meltdown mitigation enabled it was adding a 2-3 sec pause to simply loading the start menu in a windows 10 guest, disabling the mitigation took it down to about 1 second, so the overheads were clearly large although thats an older chip that doesnt have PCID enhancements (its instant like bare metal on my 2600X). When people talk about it only impacting i/o they assume if you not doing i/o to a storage device then its no impact, but ram access is also i/o. Cpu's are constantly doing i/o to ram.
For hypervisor workloads the performance impact on intel chips is huge, if you enable every possible mitigation you can wave bye bye to circa 50% of your cpu grunt on an intel platform, I am not joking, in that case a 2.5ghz AMD chip could match a 5ghz intel chip.
There is many youtube videos showing gimped intel cpu's to show matched clock to clock performance.
I didnt agree with those videos as they were gimping intel, but they show info you want.
For me its per core performance rather than per clock performance that matters.
I didnt agree with those videos as they were gimping intel, but they show info you want.
For me its per core performance rather than per clock performance that matters.
Why gimped, its a fair comparison to show performance clock for clock between the two.
If AMD are matching Intel at the same clock rate with the 14nm units then 7nm should show some nice gains
As mentioned above though, if AMD released a chip that ran 4GHz and matched the performance of an Intel at 5Ghz, people would still say the Intel is better, when in actual fact the 4GHz chip will be cooler, cheaper to run and potentially last longer.......
If AMD are matching Intel at the same clock rate with the 14nm units then 7nm should show some nice gains
As mentioned above though, if AMD released a chip that ran 4GHz and matched the performance of an Intel at 5Ghz, people would still say the Intel is better, when in actual fact the 4GHz chip will be cooler, cheaper to run and potentially last longer.......
Why gimped, its a fair comparison to show performance clock for clock between the two.
If AMD are matching Intel at the same clock rate with the 14nm units then 7nm should show some nice gains
As mentioned above though, if AMD released a chip that ran 4GHz and matched the performance of an Intel at 5Ghz, people would still say the Intel is better, when in actual fact the 4GHz chip will be cooler, cheaper to run and potentially last longer.......
Yes and thats why i say it should be per core not per clock.
If a 4ghz AMD runs as well as a 5ghz intel then its the same per core performance which is ultimately what matters not per clock.
All these clock for clock tests achieved by gimping one of the two cpus been tested wrongfully puts an over emphasis on per clock performance. Sadly most of the testing done during the ryzen 2 launch for whatever reasons people had were been done to try and make the ryzen 2 chip look as good as possible which led to a lot of unfair testing been carried out, intel pricing is insane, their 9 series chips are poor, it didnt need bad testing to point that out.