This is a NUCLEAR BOMB !!! Ryzen 2600 gets 17% single core performance boost

R

Rakanoth

n00b
Joined
Oct 6, 2017
Messages
48
As you guys know, AMD is to release the Ryzen 2000 or let's just call it Zen+ In April. In January the Ryzen 5 2600 already surfaced in the SiSoft Sandra database. The entry showed a processor called Ryzen 5 2600, which obviously is Zen+, the model listed is a six-core twelve threaded processor.

The very same processor once again has surfaced and seems to be the counterpart of the Ryzen 5 1600. The Ryzen 5 1600 shows Family 23 Model 1 Stepping 1, the new Ryzen 2600 reads out as Family 23 Model 8 Stepping 2.

The entry within the SiSoft database is:

ZD2600BBM68AF_38/34_Y (6C 12T 3.4GHz, 1.1GHz IMC, 6x 512kB L2, 2x 8MB L3).

And yes you can deduct anything and pretty much everything from that, including a 3.4 GHz base clock and a 3.8 GHz turbo. The new leak was spotted at GeekBench and seem to be interesting as it shows that exact same product code, the single core score returns a 4269 points and the multicore score now sees 20102 points. So let's call that a ~10% performance increase. Obviously, we will still need to learn if the new Ryzen 2000 series can tweak higher compared to its predecessor.

Now I know, it's not much to look at and Geekbench most definitely is not definitive and all saying, but I made the following comparison. BTW I assume the Ryzen 2600 is a non 'X' model much like a Ryzen 1600, and as such compared based on that being the fairest baseline. Note, the L 1/2/3 caches on the original Ryzen 5 1600 entry seems to be messed up, for the Ryzen 5 2600 they seem spot on. Also, the Insyde Software BIOS for Ryzen 5 2600 might seem weird, but that is a UEFI Firmware & Engineering Services Provider, indicating this test was using an engineering sample motherboard.
Click to expand...

Source: http://www.guru3d.com/news_story/amd_ryzen_2600_benchmark_spotted.html

Please look at the images at the link. I could not copy the images here as I was rushing to post this awesome news.

Old Ryzen 1600 score: 3636
New Ryzen 2600 score: 4269 !!!!!

The boost: 17% !!! Wow wow wow
 
1600's are 3.2/3.6 turbo.

That 2600 is 3.4/3.8 turbo.

I dont know how a higher clocked cpu scoring higher is news. Its just boring logic.

What we are all waiting for is seeing what, if any improvements there are in OC headroom, supported memory, and IPC changes. In that order.

PS: I want the the ryzen 2x00's to be awesome. This is not proof.
 
True they did up the clocks, but 200mhz doesnt = 17% performance, Maybe 5-7% if that?

Either way it looks like IPC was increased a lil bit. Maybe 8-10%. Also we have no idea how much further Zen+ can overclock. Hopefully at LEAST 200mhz.

just 0.02c
 
This is probably due to the way the new turbo clocks work, with the chip spending more time towards the higher clocks under full load.

I'm more excited about what this means for the 2600X and bigger chips. If the non-x is hitting 3.8, then it begs the question of how AMD plans to make the X worth the purchase. If my hunch is right, then AMD will want to make the X series chips more valuable and boost the clocks substantially.
 
I wouldn't call it a nuclear bomb, but >10% increase in both single and multi core performance is pretty good. I want to see what the top of line X parts can do.
 
Nucular bomb? Even GW Bush would say that is silly.

Trust me, I want AMD to succeed. But we should just relax and see what happens.

I remember how much fun I had OC'n my Barton 2500+. I'm not so sure that we'll see that kind of stuff any time soon with AMD, but I'll gladly go back to them.

If nothing else AMD is forcing Intel to stop giving us small incremental upgrades.
 
agreed, nuclear bomb? maybe more of a targeted missile strike. It's not bad news, but at the same time there is still too much unknown.
 
This is not bad at all, 5% faster for single core and 5% slower on multi-core side vs 1700X.

Hope they left some headroom for X versions.
 
guys complaining, it is still a 10.5% increase in performance clock for clock.
 
mnewxcv said:
guys complaining, it is still a 10.5% increase in performance clock for clock.
Click to expand...
Isn't that not what the process was suppose to achieve? A 10% increase?

In any case if this holds up it is a bigger improvement then what Intel has been showing for awhile from one generation to the next. I too am more interested in the OC headroom these new processors will show.
 
Geekbench has the r5 1600 at 3900 points elsewhere so not that reliable of a benchmark.

OP, could you cut back on the energy drinks? This is not that amazing.
 
Nightfire said:
Geekbench has the r5 1600 at 3900 points elsewhere so not that reliable of a benchmark.

OP, could you cut back on the energy drinks? This is not that amazing.
Click to expand...

OP be like

latest
 
It's a nice bit of news, but until we see real world benchmarks, and have a better understanding of how this performance increase is being achieved (higher XFR maybe? Or more time in boost?) we won't know what this really means.

Between clockspeed increase, some tweaks to boost clocks, maybe some little bit of IPC, and some memory controller tweaks, 10-15% overall gain is not unreasonable. But I'd want to see more than one single benchmark before I took that to the bank.
 
trick0502 said:
4.6+ oc would be a nuclear bomb
Click to expand...

Yeah, that would get me to upgrade from my 4ghz FX-6300. For funzies i ran that geekbench 4.2 and got 2657 / 8988 on my FX - 6300. A much lower score, but you wouldnt know it by using it. Runs just as well as my ryzen 1700 for my general use.
 
Last edited:
Hate to be that guy but... is this supposed to be impressive or even respectable? My laptop's frequency-limited, 2.5 year old 6700HQ using single-channel, low-speed DDR4 has as high of a single-core score in Geekbench as that 2600 (and... substantially better than a stock 1600? o_O). Regardless of what the benchmark does or doesn't represent (and there's a whole lot it probably doesn't), that's a pretty unimpressive score for a 2018 desktop core (especially if Zen+ is limited to a similar max frequency range as Zen).

This is what a nuclear bomb looks like:
Geekbench 8700K Cropped.png

>50% better single-core score (stock should be about 6% lower than that score given the 4.7 GHz single-core turbo, but the point remains: Ryzen's competition apparently dominates in Geekbench); I'm with the others that need some other proof of greatness because I think Geekbench only actually works against Ryzen's presentation.
 
Last edited:
I think the point was that there is a 17% increase, not that Geekbench is relevant to anything. Realistically, would I rather compute on a daily basis with your 6700HQ or a Ryzen 1600/2600? I'd probably take the Ryzen regardless of what Geekbench says.
 
trick0502 said:
4.6+ oc would be a nuclear bomb
Click to expand...
would be hilarious to see what would happen in intel camp then lol :p but hey, even then it would be hard to beat intel i think. but it is likely they spy on eachother so they probably have some clue what they are doing, even tough that is illegal.
 
A 17% increase in the context of a completely unrepresentative benchmark means nothing though. I mean, if there's a 17% increase here that means anything, yet Coffee Lake curbstomps the end result even harder than usual despite it, why should anyone care? On the other hand, if it is irrelevant as you say (and I agree), that 17% gain also tells us nothing of applicable worth (most certainly nothing about generalized gains as some suggest). Really can't have it both ways, which I intended to be the point of my post.

Either the benchmark is relevant and the score is honestly more impressively weak than anything or it's not relevant, but that 17% gain could come from the most minor and inconsequential of improvements that real-world software won't reflect at all (that GB just so happens to stress). I lean towards the latter, but still, scores can't be analyzed in a vacuum to determine impressive gains.

I think I saw that even the A11 in the latest iphone matches this single-core score somewhere. Geekbench is funny.
 
Last edited:
but a11 is quite a few cores and 2 of them, i just looked at wiki runs at 2.39 ghz? and they got 10nm finfet. so combining all these cores together it makes sense that it could beat out 1 amd core, unless it was 1 core vs 1 then seems odd :p it's just crazy how phones are these days lol... but yeah, real world and proper testing is what im waiting for anyway. cause i do remember the 1080 hypeing even with reviews o_o and u had to look to find a non BS review. cause in my mind an upgrade it goes from the last product in line, not backtracks, so compare to 980 maybe it was good, but it sure as shit wasnt worth replacing 980ti over as alot of ppl wanted u to believe. it still probs wont be enough or wort for me to upgrade cpu. im probably getting next ti before.
 
Sure it mean something when you are comparing Apples to Apples like that. 17% sounds like the expectation. So essentially it is just confirmation that it's performance is about where it was expected.

I still wouldn't take your 6700 over it though :p
 
Lol. I understand why you thought I meant the combined score, but I actually meant single-core vs single-core. The A11's single-core score is literally like 4200 itself. We live in this bizarre universe, apparently:

Geekbench-A11-Bionic-image-001.png

kirbyrj said:
Sure it mean something when you are comparing Apples to Apples like that. 17% sounds like the expectation. So essentially it is just confirmation that it's performance is about where it was expected.

I still wouldn't take your 6700 over it though :p
Click to expand...

But it's not Apples to Apples (not like taking the same Zen CPU and merely overclocking it). Zen+ is a different CPU lineup based on a different chip. It is very likely to have (relatively minor) changes, but even relatively minor changes can be exploited by unrealistic benchmarks like GB to make those seem like a big deal. Even firmware and microcode can be used to "cheat" on certain benchmarks while doing little for real applications.
 
Ghost of Cyrix said:
But it's not Apples to Apples (not like taking the same Zen CPU and merely overclocking it). Zen+ is a different CPU lineup based on a different chip. It is very likely to have (relatively minor) changes, but even relatively minor changes can be exploited by unrealistic benchmarks like GB to make those seem like a big deal. Even firmware and microcode can be used to "cheat" on certain benchmarks while doing little for real applications.
Click to expand...
I don't think it's that much different that it will make a big difference. If release reviews show various other benchs with that same 12-18% difference, I guess I was right. If it's somewhere outside of that I was wrong.
 
I will be pleasantly surprised if get 4.4ghz stable overclocks out of these new Ryzen chips. If we do I think even that will be a pretty big step forward. I think more realistically we'll see higher stock clocks and some very slightly increased o/c headroom (4.1-4.2ghz on more chips than you see now). But that's just my guess :p
 
I am not responding to the truth on the rumour but I am going to argue that 17% if accepted as correct for sake of argument, it is a very big gain in a single generation. Kyle did his own benches here overclocking a Kabylake and Sandy to 4.7ghz and the Kabylake didn't even pass 20% performance improvement over the old sandybridge, to do it in a single generation with a basic 200mhz bump is somewhat superhero like and that is where the belief in it comes, given that Intel claimed 15% over skylake with the kaby release but in proviso was the very fine print that the 7700K was compared against a 6600 and the clocks are 800mhz apart, for AMD to do 17% in 200mhz is a feat of brilliance.
 
Ghost of Cyrix said:
A 17% increase in the context of a completely unrepresentative benchmark means nothing though. I mean, if there's a 17% increase here that means anything, yet Coffee Lake curbstomps the end result even harder than usual despite it, why should anyone care? On the other hand, if it is irrelevant as you say (and I agree), that 17% gain also tells us nothing of applicable worth (most certainly nothing about generalized gains as some suggest). Really can't have it both ways, which I intended to be the point of my post.

Either the benchmark is relevant and the score is honestly more impressively weak than anything or it's not relevant, but that 17% gain could come from the most minor and inconsequential of improvements that real-world software won't reflect at all (that GB just so happens to stress). I lean towards the latter, but still, scores can't be analyzed in a vacuum to determine impressive gains.

I think I saw that even the A11 in the latest iphone matches this single-core score somewhere. Geekbench is funny.
Click to expand...

Geekbench has become a very popular Intel showcase bench suite, making its way onto intel slide shows, I remember when it was Cinebench and Civilizations for years on end. The other one is Intel In house sysmark bench.
 
Ghost of Cyrix said:
Hate to be that guy but... is this supposed to be impressive or even respectable? My laptop's frequency-limited, 2.5 year old 6700HQ using single-channel, low-speed DDR4 has as high of a single-core score in Geekbench as that 2600 (and... substantially better than a stock 1600? o_O). Regardless of what the benchmark does or doesn't represent (and there's a whole lot it probably doesn't), that's a pretty unimpressive score for a 2018 desktop core (especially if Zen+ is limited to a similar max frequency range as Zen).

This is what a nuclear bomb looks like:
View attachment 54880

>50% better single-core score (stock should be about 6% lower than that score given the 4.7 GHz single-core turbo, but the point remains: Ryzen's competition apparently dominates in Geekbench); I'm with the others that need some other proof of greatness because I think Geekbench only actually works against Ryzen's presentation.
Click to expand...

Yip Geekbench has for the last year and a bit become a Intel slideshow benchmark, it is now a completely unrepresentative product which most reviewers just drop. It reeks of the ol "genuine intel" fix that cinebench used to have in their code to boost intel scores. It would be nice if this was Cinebench rather.
 
Before claiming nuclear "17% single core boost" and "10.5% increase in performance clock for clock" I would perform some basic tests, such as checking if the scores correspond to the same version of benchmark, and so...

I go to the OP link and I can read in the image one score corresponds to 4.1.4 version whereas other is for 4.0.3 version

index.php


I can also see the score reported for the R5 1600 is too low compared to scores obtained by other 1600 chips; the 1600 system must be somewhat broken because it is reporting incorrectly the caches on the system. Here we have another 1600 scoring 4263 and 19453 on the version 4.1.1 of the benchmark, and with all caches correctly reported.
 
to much in that leak to suggest not taking it seriously, it is also an odd bench for AMD to use given how Geekbench is not used in benching test suites. If someone was going to show promising results they would go Cinebench, Blender
 
buttons said:
Yeah, that would get me to upgrade from my 4ghz FX-6300. For funzies i ran that geekbench 4.2 and got 2657 / 8988 on my FX - 6300. A much lower score, but you wouldnt know it by using it. Runs just as well as my ryzen 1700 for my general use.
Click to expand...

I know what you mean. This is why I wish I had stuck with my FX 8350 work computer over upgrading it to a Ryzen 7 1700. I just rebuilt a computer with an FX 8350 system for a customer with a new 250GB SSD and with Windows 10 Home, that this is really fast. No, not benchmark fast but feel and do things fast.
 
You must log in or register to reply here.
Back
Top