Silicon Lottery Announces Ryzen 3000 Bins

[leadeater]

5 minutes ago, cj09beira said:

i think good data on 3800x is needed for us to know what is really going on, a single core Fmax-Vmin vs all core would be quite useful

GN has a review of the 3800X, the summary of it was don't waste your money. Once you remove/increase the PPT limit it's solely down to the die quality, which his 3800X appears to be worse than his 3700X. In spec operation did have the 3800X doing slightly better but that's really just the higher PPT allowing slightly better clocks, you already know that anyway .

 

I wouldn't buy a 3800X knowing this.

[CiBi]

I bought a 3800X before GN did the video on it. Here's hoping its a better bin

[Dabombinable]

And here I am, one of the people who'd rather take a Pentium4 (631 in my case) for a spin, and overclock it well past 5GHz, instead of worrying about getting the same out of a far newer architecture...which isn't designed to clock high at all costs.

I still love that old lofty goal of 10GHz at stock...

[cj09beira]

2 minutes ago, Dabombinable said:

And here I am, one of the people who'd rather take a Pentium4 (631 in my case) for a spin, and overclock it well past 5GHz, instead of worrying about getting the same out of a far newer architecture...which isn't designed to clock high at all costs.

I still love that old lofty goal of 10GHz at stock...

is that on air? i had a p4 3.2ghz for socket 775

[Dabombinable]

Just now, cj09beira said:

is that on air?

Pentium 4 631, it can do that on air (65nm with a decently soldered IHS).

Well, it used to before the P5Q Deluxe went "senile" and decided that it won't run my good DDR2 kit in dual channel.

[Drak3]

10 minutes ago, cj09beira said:

thats what makes it a clock wall not that there is a max safe voltage point that always exists.

Ryzen 3000 takes the same effort to push past 4.3 that Ryzen 2 did 4.2 and Ryzen 1 did 4.

 

If you're going to keep saying that Ryzen 3 isn't hitting a clock wall, do the same for the other chips that give the same reactions in the same scenarios. Until then, your argument is self contradictory.

[Princess Luna]

26 minutes ago, CiBi said:

I bought a 3800X before GN did the video on it. Here's hoping its a better bin

Hardly... if anything both Tech Deals and Hardware Unboxed got higher clocks out of the R5 3600 sample than they did from the R5 3600X sample... so I personally believe the X does not mean "higher binning / purer silicon" what so ever.

[cj09beira]

2 minutes ago, Drak3 said:

Ryzen 3000 takes the same effort to push past 4.3 that Ryzen 2 did 4.2 and Ryzen 1 did 4.

 

If you're going to keep saying that Ryzen 3 isn't hitting a clock wall, do the same for the other chips that give the same reactions in the same scenarios. Until then, your argument is self contradictory.

i don't have one to test myself, if there was this test done already and you seen it i would gladly add it, derbauer almost did it but he locked the wrong variable .

about if zen 2 as a whole has a clock wall or not i am not 100% sure, because of 2 things, one is that the fact that in single core they can clock significantly higher than all core, its something that had not happened for some time, 100mhz higher sure but 200-300 seems a bit high, then you have the chiplets adding another variable to the equation, which affects the distribution of the good silicon, and i think many reviewers in the attempt to show frequency gains set voltages way to high which changes the numbers 

[Drak3]

Just now, cj09beira said:

one is that the fact that in single core they can clock significantly higher than all core

Again, we're not seeing that happrn outside a few golden chips. Turbo clocks are not, and never have been, guaranteed. They're just a 'the chip may do this under the right scenario.

[yolosnail]

I don't get why everyone is so hung up on the clock speed, if Intel released a new CPU that couldn't hit 5GHz but matched or exceeded the 9900K at 5GHz with lower power draw, everyone would be praising Intel's achievements.

 

Let's not forget AMD had 5GHz over half a decade ago, but with pitiful IPC, so Intel's chips that were clocked lower performed better

Now that we have Zen 2 with IPC that is ahead of Intel, everyone is suddenly outraged that AMD haven't hit 5GHz

 

At the end of the day, the only people interested in the clock speed are those who are ahead, what 'regular' people should really be focusing on is the performance

[WkdPaul]

* Thread cleaned *

 

Let's keep the topic civil.

[S w a t s o n]

3 hours ago, leadeater said:

Yes which was exactly the point, same node, exactly the same node, with only an architecture tweak and gained 500Mhz single core boost. That's actually pretty good honestly. The problem with the 5GHz expectation was that equated to a 700Mhz increase which is right in the unrealistic area of expectation. Much of the anticipated gains people were expecting was from the shift from GF to TSMC and 7nm, evidence shows these types of moves and shrinks does not equate to large clocks gains. The majority of clock gains have come from architecture improvements, ones targeted to increase clocks.

 

Zen2 had to counteract both an expected clock reduction from 7nm due to internal resistance increase and a higher power target per core due to increased FPU bit width, two changes that inhibit increasing clocks. AMD and TSMC did an excellent job in being able to increase the clocks, very excellent job.

 

That doesn't change this issue that these points were raised before Zen2 launch and continuously brushed off, as well as the clock trend point. Why expect the unrealistic, why hope for something so unlikely? Why as it got closer you have some reviewers outright say 5Ghz is not happening prefer to believe 5Ghz rumors when that is the single piece of information that would back that against multiple others pointing towards it not being a thing.

 

Hindsight is a thing, but honestly hindsight was not required to know 5Ghz was so unlikely to happen it really wasn't worth talking like it was a possibility at that stage.

 

I only use that term when others use to it address that fact it was used.

 

Zen1 and Zen+ didn't have a clock wall either, both can clock higher with LN2 just the same as any other CPU. Zen2 won't clock higher with increased vcore, it's not about unsafe or the temperature caused by the vcore increase it's about the chemical properties and how they act at lower temperatures. You can bring the CPU down to say 10C and no matter how much you pump the vcore up and leave it consistently at 10C while doing it the maximum attainable clock will not increase, lower temperature is required.

 

There is a point where increasing vcore will net you nothing and is entirely temperature dependent. If you've watched a number of LN2 overclocking videos you would have seen this in action whether you took much note of it or not.

I am aware of all the technical hurdles AMD faced, i mentioned some in my post. Yes changing fpu width doesn't help but zen 2 doesn't need or really have much of an avx2 offset compared to Intel who does. Yes they very clearly had a clock wall. Let's just stop talking about clock wall if people can't seem to understand it.

 

The increases did not come purely from architecture for devils Canyon, same node doesn't mean anything. 14++ isn't 14+ isn't 14. Each one improved on the other

[thorhammerz]

4 hours ago, cj09beira said:

as i expected the chiplet design is having a major effect on overclocking headroom for the lower end chips, though i did not expect amd to go ahead and put one slow one even on the 3900x, they must be really needing those good ones for epyc, 

Don't forget about that special binning for the upcoming 3950x .

 

They'd also probably need a fair of the good dies for TR 3000 (if/when that comes out), especially if those functionally end up being higher-clocked Epycs.

[melete]

5 hours ago, Drak3 said:

Ryzen 3000 takes the same effort to push past 4.3 that Ryzen 2 did 4.2 and Ryzen 1 did 4.

 

If you're going to keep saying that Ryzen 3 isn't hitting a clock wall, do the same for the other chips that give the same reactions in the same scenarios. Until then, your argument is self contradictory.

Single and two core boosts go past 4.3GHz just fine on air. 4.4GHz is common for me, and perhaps if I had a better AGESA revision and a different motherboard like an X570 I’d see 4.5GHz. No complicated cooling necessary. My 3700X won’t do beyond 4.3GHz at sane voltage in an all-core OC, sure, but it’s worth noting that the way these chips boost is a little different than other processors. 

[Derangel]

23 minutes ago, melete said:

Single and two core boosts go past 4.3GHz just fine on air. 4.4GHz is common for me, and perhaps if I had a better AGESA revision and a different motherboard like an X570 I’d see 4.5GHz. No complicated cooling necessary. My 3700X won’t do beyond 4.3GHz at sane voltage in an all-core OC, sure, but it’s worth noting that the way these chips boost is a little different than other processors. 

As long as you have some airflow over the VRMs, you're not likely being limited by the motherboard itself. While ASUS had some notoriously gimped VRM implementations on their X470 boards they're still more than enough for the 3700X.

[melete]

9 minutes ago, Derangel said:

As long as you have some airflow over the VRMs, you're not likely being limited by the motherboard itself. While ASUS had some notoriously gimped VRM implementations on their X470 boards they're still more than enough for the 3700X.

My particular motherboard has a 6 phase with IR3555 60A powerstages, so it’s definitely not limited by VRMs on a 3700X. The limitations I’m referring to is more about the BIOS and especially the AGESA revision the motherboard is using.

[mr moose]

7 hours ago, yolosnail said:

I don't get why everyone is so hung up on the clock speed, if Intel released a new CPU that couldn't hit 5GHz but matched or exceeded the 9900K at 5GHz with lower power draw, everyone would be praising Intel's achievements.

 

I beg to differ, I think many would be arguing that Intel lied on roadmaps and were sitting on this lower power product being all anti consumer and anti trust.

[leadeater]

6 hours ago, S w a t s o n said:

The increases did not come purely from architecture for devils Canyon, same node doesn't mean anything. 14++ isn't 14+ isn't 14. Each one improved on the other

Devil's Canyon used the same node, it wasn't another + and it was on 22nm anyway. The clock increase came from an arch tweak, new paste material and an increase in package TDP.

[S w a t s o n]

3 hours ago, leadeater said:

Devil's Canyon used the same node, it wasn't another + and it was on 22nm anyway. The clock increase came from an arch tweak, new paste material and an increase in package TDP.

Just because intel didnt name it 22nm++ doesnt mean it wasnt improved over say a 3770k at the OG 22nm process

[MMKing]

12 minutes ago, S w a t s o n said:

Just because intel didnt name it 22nm++ doesnt mean it wasnt improved over say a 3770k at the OG 22nm process

No it doesn't, you are correct. But Devil's Canyon (i5-4690K and i7-4790K) being made on a different 22nm process than either the other Haswell series CPUs or the Ivy bridge 3000 series CPUs is not the default position.

[S w a t s o n]

Just now, MMKing said:

No it doesn't, you are correct. But Devil's Canyon (i5-4690K and i7-4790K) being made on a different 22nm process than either the other Haswell series CPUs or the Ivy bridge 3000 series CPUs is not the default position.

I would say it is the default position. Intel is not going to not tweak the process when they can even across a given generation that is already in HVM. Besides trying to improve yields

[MMKing]

Just now, S w a t s o n said:

I would say it is the default position. Intel is not going to not tweak the process when they can even across a given generation that is already in HVM. Besides trying to improve yields

Intel was still in the mindset of a ''tick tock'' cycle. Where Tick was the process improvement going from 32nm (sandy bridge 2000 series) to 22nm (ivy bridge 3000 series). While Tock was the micro architecture improvements within a process. Going from the 3000 series to the 4000 series, there were no noteworthy changes to the 22nm process. At least none worth marketing. Why? Because the Haswell 4000 series was released June 2013, while the Broadwell (5000) 14nm was released in October 2014. With the actual successor for the mainstream market coming August 2015. Shortly after, around March 2016, Intel abandonen the ''tick tock'' scheme.

 

Prior to the abandoning of the ''tick tock'' idea. There were no significant improvements made to existing processes. Rather, the improvements was a transistor shrink. Why allocate resources improving the 2012 22nm process, when the 14nm process is expected in 2014?

[S w a t s o n]

1 minute ago, MMKing said:

Intel was still in the mindset of a ''tick tock'' cycle. Where Tick was the process improvement going from 32nm (sandy bridge 2000 series) to 22nm (ivy bridge 3000 series). While Tock was the micro architecture improvements within a process. Going from the 3000 series to the 4000 series, there were no noteworthy changes to the 22nm process. At least none worth marketing. Why? Because the Haswell 4000 series was released June 2013, while the Broadwell (5000) 14nm was released in October 2014. With the actual successor for the mainstream market coming August 2015. Shortly after, around March 2016, Intel abandonen the ''tick tock'' scheme.

 

Prior to the abandoning of the ''tick tock'' idea. There were no significant improvements made to existing processes. Rather, the improvements was a transistor shrink. Why allocate resources improving the 2012 22nm process, when the 14nm process is expected in 2014?

Tick Tock was always a marketing scheme. I refuse to believe intel's foundry wasnt making constant tweaks to each node during it's lifetime rather than waiting for the next node.

[MMKing]

None worth marketing. None worth even mentioning. The 14nm+ and the 14nm++ represents a significant divergence from the original 14nm process. Haswell 22nm does not represent a significant divergence from Ivy Bridge 22nm

 

Look at the goddamn product pages

3770k - https://ark.intel.com/content/www/us/en/ark/products/65523/intel-core-i7-3770k-processor-8m-cache-up-to-3-90-ghz.html

4770k - https://ark.intel.com/content/www/us/en/ark/products/75123/intel-core-i7-4770k-processor-8m-cache-up-to-3-90-ghz.html

Compare page - https://ark.intel.com/content/www/us/en/ark/compare.html?productIds=75123,65523

 

There is only two differences, not counting graphics differences.

 

The 4770k has a TDP 84W over 77W. And the 4770k has AVX2.0, while the 3770K has AVX1.0. A architectural improvement. The 4770k is superior, because the instruction set is superior NOT because the process is superior. Heck, there are even some tests where the 3770K performs better than the 4770k. But even where the 4770k scores better, the difference is insignificant.

 

 

You can call Tick Tock a marketing scheme all you want. It was a marketing scheme rooted in reality.

32nm

2009/2010 - Nehalem

2011 - Sandy bridge

22nm

2012 - ivy bridge

2013 - Haswell

14nm

2014 - Broadwell

2015 - Skylake

Tick tock broken, 14nm continues onward.

 

If it's so self evident. You should have no issues producing some evidence to support that there is any difference at all between the 22nm silicon found in the 3000 series, and the 22nm silicon found in the 4000 series.

[S w a t s o n]

6 minutes ago, MMKing said:

None worth marketing. None worth even mentioning. The 14nm+ and the 14nm++ represents a significant divergence from the original 14nm process. Haswell 22nm does not represent a significant divergence from Ivy Bridge 22nm

 

Look at the goddamn product pages

3770k - https://ark.intel.com/content/www/us/en/ark/products/65523/intel-core-i7-3770k-processor-8m-cache-up-to-3-90-ghz.html

4770k - https://ark.intel.com/content/www/us/en/ark/products/75123/intel-core-i7-4770k-processor-8m-cache-up-to-3-90-ghz.html

Compare page - https://ark.intel.com/content/www/us/en/ark/compare.html?productIds=75123,65523

 

There is only two differences, not counting graphics differences.

 

The 4770k has a TDP 84W over 77W. And the 4770k has AVX2.0, while the 3770K has AVX1.0. A architectural improvement. The 4770k is superior, because the instruction set is superior NOT because the process is superior. Heck, there are even some tests where the 3770K performs better than the 4770k. But even where the 4770k scores better, the difference is insignificant.

 

 

You can call Tick Tock a marketing scheme all you want. It was a marketing scheme rooted in reality.

32nm

2009/2010 - Nehalem

2011 - Sandy bridge

22nm

2012 - ivy bridge

2013 - Haswell

14nm

2014 - Broadwell

2015 - Skylake

Tick tock broken, 14nm continues onward.

 

If it's so self evident. You should have no issues producing some evidence to support that there is any difference at all between the 22nm silicon found in the 3000 series, and the 22nm silicon found in the 4000 series.

What? So because intel doesnt market the nitty gritty of any and all improvements made on the foundry side that means anything? I dont see how literally anything you said shows otherwise. 4770k having new features doesnt mean the process node was exactly the same. If the 3770k wins on any test it's probably because of the IVR if anything.

 

Just because they stepped up the marketing after they failed at tick-tock and just went to Arch-Process-Optimize doesnt mean they werent optimizing before. 14+ marketing exists because of the optimize step in the marketing, which as you pointed out wasnt so much a thing when intel just marketed it as tick then tock.