Clouseau:

Which bios version is being used? Supposedly the cold boot bug was put to bed since version 3008. Since then version 1501, 1502, and currently 6001 has been released and have not read any new or continuing issues regard it. Only means I have not read anything about it since. Why I am surprised it still is an issue. RAM compatibility supposedly has improved as well. For me anything since version 3008 has been a no go. Compared to my M3A32-MVP this board has been a dream. Then again that has just been my experience. Personally will be holding out till the following series is released or 3600 ram speed has been confirmed by more than a select few with a "golden" chip.

helpful post, liked. 3600 sounds very cool for an apu, even if only 8GB. Important, little noted development is hardware unboxed's tests show 64MB reserved memory for the apu's gpu works as well as reserved memory - ie. allowing windows? to dynamically allocate memory to the gpu works well. If this finding holds, this makes 8GB doable for APUs.

Slight clock speed boost and memory speed improvement would make Ryzen2 a great gaming CPU. Hopefully Ryzen2 will shake Intel once more.

fl2015:

Here's what I get on my 1600 dropped down to 3.8 ghz https://browser.geekbench.com/v4/cpu/7106573, if the 2600 uses a 3.8ghz single core turbo boost then the single core performance doesn't look too impressive.

Did you also downclock your RAM to 2.6/2.9? AMD are likely only testing using un-OC'd RAM for their tests.

AlmondMan:

Did you also downclock your RAM to 2.6/2.9? AMD are likely only testing using un-OC'd RAM for their tests.

It's at 2933mhz but that's not exactly going to make a huge difference, point is single core IPC is around the same while multicore performance is a tiny bit better. The big question is whether the improvements have translated over to max achievable clocks, however those expecting 4.4/4.5 ghz are probably living in dreamland.

mikev190:

But next you'll be telling us there's no point upgrading from an i7 2600k.

Mmm, hard to call anything as an “good upgrade” when comparing to 5GHz 2600K for gaming. (yet) Perhaps time and ZEN2 will change that.

Anyone putting ANY faith in a single Geekbench (yikes!!) score on a Ryzen chip no less (Zen plays with GB even more wonkiely than Intel's chips do, and they're all over the freaking place) to compare performance between this chip and anything else is pretty much just wasting time. Or hell, use Geekbench as anything but the most ROUGH (judgements like say the chip is a 4c/8t i7). You can run Geekbench with a 6700K & 7700K & have the former come out on top regularly, and that's not even the weirdest results it can spit out. By and large it's a worthless benchmark, so I wouldn't put ANY stock into the scores vs say, your R5 1600. Pretty much the only valuable info here are the clock speeds, which are far from guaranteed to be final. Even this close to the Ryzen 7 launch SKU's announcement, late Zen 1st Gen engineering samples with close, but slightly reduced clocks (far higher than the 1st samples spotted, but a bit lower than final; maybe to avoid spoiling final clocks *shrug*) were also popping up so my guess has just a good chance of being 100-200MHz higher in it's retail form as it's clocked here. But the most important part is THESE SCORES ARE TOTALLY WORTHLESS!!!

sverek:

Mmm, hard to call anything as an “good upgrade” when comparing to 5GHz 2600K for gaming. (yet) Perhaps time and ZEN2 will change that.

That's not really true any more finally (and thank goodness! As in we're finally making significant ground past Sandy, not that Sandy's legendary longevity winding to a close is a happy thing haha). It's very easy to bottleneck a cranked 2600K significantly with the very latest games @ 1080p with a fast card vs cranked Ryzen 5/7 for a big performance disparity (with the Ryzen's falling above Sandy/Ivy, but beneath Skylake/KL/CL ofc). Granted these are all well above 60fps figures, so if that's your target or you have a mid-range card ya got nothing to worry about for a good bit more, but running 144Hz or with today's high end cards is really stretching the bridge chips now (lol) in many games, and it's only getting more and more common. Is the former a bad gaming experience yet? Far from it, but with the core wars & serious market competition here again, it's days have been numbered (fina-freaking-ly!!! Took damn long enough!!! Hahahaha)

mikev190:

Not necessarily an amazing Ryzen 1 can get 4.1ghz so 4.4ghz could be doable. Also this single core nonsense is becoming less important. All modern games coming out use 4 threads+ so unless you play wow or some other old game then I wouldn't bother worrying about single core. A good example is the new vermintide game uses 13-14 threads on a 8core/16thread cpu. But next you'll be telling us there's no point upgrading from an i7 2600k.

Going beyond 4ghz on Ryzen 1 is extremely rare and requires an insane amount of cooling for the little performance gained. As for games using multicore, yes some might be but all the ones most people actually play as well as the 'niche' games you're likely to pick up in a steam sale or something are most likely better with stronger single core performance. Not everyone just plays the AAA flavour of the month or the tech tubers bench suite, I even got games that are 6-7 years old now I still fire up. As for 2600k to 1600, they are not night & day. Some games perform better on the the Ryzen others perform better on the 2600k, in a game I play regularly I actually experience slightly worse performance & stutters on Ryzen due to the engine not being optimized for it. Games like BF1 & TW3 are better on the 1600 though I didn't sidegrade for a performance boost but just to move to a newer platform for features like NvMe, PCI-e 3.0, RGB etc as well as the fact I only paid £136 for my 1600 on sale which i made a good chunk of that back selling the 2600k.

from the shots on the Original G3d article above, its seems they swapped the L1 ins and L1 Data cache sizes, and doubled the L3 . Wouldn't this affect execution speed positively?

Clouseau:

Which bios version is being used? Supposedly the cold boot bug was put to bed since version 3008. Since then version 1501, 1502, and currently 6001 has been released and have not read any new or continuing issues regard it. Only means I have not read anything about it since. Why I am surprised it still is an issue. RAM compatibility supposedly has improved as well. For me anything since version 3008 has been a no go. Compared to my M3A32-MVP this board has been a dream. Then again that has just been my experience. Personally will be holding out till the following series is released or 3600 ram speed has been confirmed by more than a select few with a "golden" chip.

Since 3008 as in it was in 3008? Because i'm on 3008 currently.

Nebula:

So were looking at 5-13% IPC increase and for all of us better overclockability, so with 4,2-4,4GHz we should look at about 8-17% performance increase with lower/or same wattage.

Unless I'm missing something or not understanding you correctly, where are you finding evidence that these new chips will OC better? I hope you're right, but I haven't heard any evidence that they will.

Aura89:

No matter which bios i try, no matter what memory settings i do, i have had a constant issue with cold boot many times deciding to not boot the OS, of which i'll have to reset the PC, or shut it fully down again, and then it'll, hopefully, maybe boot up.

I have a Biostar board and used to have the same problem as you. When I first got the board (before the AGESA 1006 updates) my 3GHz RAM was limited to 2666MHz. After that update, I can now run my RAM at 3.2GHz. But, there are a lot of settings I have to leave un-touched or else my system becomes unbootable. The weird thing is, even if I manually set the values to their defaults, the system fails to boot. For example, if I changed the tCL from "auto" to "15" (my RAM's default), it would fail to boot. But if I leave it at auto, it will run at 15. After all the updates, the only things I can change without boot failure is the frequency, the voltage (which oddly doesn't automatically adjust to match the XMP), and the command rate. Anyway my point is Biostar is a garbage company compared to Asus, so you should be able to get your RAM at it's advertised frequencies. But maybe your situation can be fixed if you approach it the way I approached mine - keep as many of the RAM settings at default as possible, and start out only adjusting the frequency, voltage, and command rate. I hear disabling GearDown helps improve stability for people, but for me touching that setting caused boot failure, whether I set it to enabled or disabled.

Aura89:

Since 3008 as in it was in 3008? Because i'm on 3008 currently.

Yes. According to Elmor's notes on the bios version the cold boot bug was addressed. That was the reason for the longer post time for 3008. Others have had good luck with 1502. Hate that my rig keeps kicking me back to version 3008. That thread over at overclock is so long now, otherwise I would have linked Elmor's post from there.

It seems to be across the net that people are calling this Zen+ as some sort of upgraded from previous Zen when even On this thread fl2015 posted near identical results from his equally clocked 1600. In my mind all Zen are the same Zen and simply upping base clocks on a chip that runs 3.9GHz all day long isn't good enough for Zen+. 3.8 is pretty shameful marketing and 100% obtainable on current ZEN. Today I am sad... There is hope, the 2600 is an engineering sample so top end isn't known yet for the finish product. I'm concerned that if we can't get 4.4GHz on an OC for binned Zen+ (2600) and there is no real IPC gains clock for clock over 1600 and little MHz gains either... If this is the outcome then ask yourself, what do we really have here? We basically have very little to no return from an expected "up to" 25% increase in performance (14nm+ was supposed to be "up to 15%", and the swap to 12nm LP was supposed to be an additional "up to 10%" over to of 14nm+). Clock for clock right now we have near 0% increase in performance/IPC and anyone posting gains in %'s should be able to see this for themselves. So far, from a product/performance standpoint (not marketing stand point), Zen+ is not an upgraded worth any attention and buying a cheaper 1600 would be the smarter move. I hope that 2800x OC levels offer significant gains as the OC results will reveal the true potential of Zen+ and we can't put aside Marketing claims and OCe to our hearts content.

Pinscher:

It seems to be across the net that people are calling this Zen+ as some sort of upgraded from previous Zen when even On this thread fl2015 posted near identical results from his equally clocked 1600.

You do realize fl2015 is comparing to the Intel 2600K, right? That's not the Ryzen 2600.

We basically have very little to no return from an expected "up to" 25% increase in performance (14nm+ was supposed to be "up to 15%", and the swap to 12nm LP was supposed to be an additional "up to 10%" over to of 14nm+). Clock for clock right now we have near 0% increase in performance/IPC and anyone posting gains in %'s should be able to see this for themselves.

Expected from who? I don't know anyone (of importance) who predicted a 25% increase from just a single generation. Zen+ is not known to be a major revision; that's what Zen 2 is for. Zen+ is just a die shrink with some minor tweaks. Hopefully that'll result in better OC and/or IPC. If it's any consolation, to my understanding, the 2400G and 2600G are not Zen+; they're just slightly tweaked Summit Ridge chips, which is why they yielded basically no performance improvement. They're still based on 14nm, whereas "Zen+" is supposed to be 12nm.

schmidtbag:

You do realize fl2015 is comparing to the Intel 2600K, right? That's not the Ryzen 2600.

I'm not, I was comparing my 1600 to the ryzen 2 results when some guy above mentioned the 2600k which I happened to own before I switched to Ryzen. My original point was the fact that by going by these results single core IPC (i.e. clock for clock) seems much the same while SMT seems to have improved (if you factor in turbo core balancing of a stock cpu), which seems kind of the opposite to what most were wishing for with wanting strong single threaded performance to compete with intel chips. Question is have AMD listened to people wanting the higher clocks as well? Is that where most of the optimizations have gone or are we looking at another 4 ghz (in general) barrier?

schmidtbag:

You do realize fl2015 is comparing to the Intel 2600K, right? That's not the Ryzen 2600.

um, did you read the same post I read. fl2015 said, "Here's what I get on my 1600 dropped down to 3.8 GHz." This translated into plain English means fl2015 provided results from his 1600. I compared his posted results VS the leaked 2600 scores. Here is the link, Friend: https://browser.geekbench.com/v4/cpu/7106573

schmidtbag:

Expected from who? I don't know anyone (of importance) who predicted a 25% increase from just a single generation.

It's not a prediction. Guru3d did an article on Global Foundries processes and had tables that showed the difference in performance from 14FinFet to 14LPP, then from 14LPP to 12nmLP. these are numbers provdied by Global Foundries The sum of potential performance gain was 25%. Go google it yourself if you want to. I guess the 2% we are going to get falls into the "up to" category so they are technically not Liars.

schmidtbag:

Zen+ is not known to be a major revision;

The reason AMD was able to go to 12nm was because little to no changes were needed beyond what has been planned for their 14nm LP refresh as the processes only differ in small optimizations. I didn't even make any points on this subject, but it's nice to continue the conversation.

schmidtbag:

If it's any consolation, to my understanding, the 2400G and 2600G are not Zen+; they're just slightly tweaked Summit Ridge chips, which is why they yielded basically no performance improvement. They're still based on 14nm, whereas "Zen+" is supposed to be 12nm.

Who's talking about 2400G and 2600G, the article is talking about 2400/2600 which are Zen+ solutions built on 12nm. You totally lost me here. The whole point of going to 14nm LPP was to eliminate FinFet imposed performance constraints in order to lift the lid on off the GHz hard cap. The switch to 12nm was supposed to take that further, according to Global Foundries and AMD a like. Regardless of arch revisions, the process changes owe us more than a 0% clock for clock IPC increase. So far, from what Ican see, Zen+ is DOA. Why buy it when you can buy a sale priced Zen on zen+ launch? "Zen+, nothing Zen can't do at a lower price point."

Pinscher:

um, did you read the same post I read.

Apparently not, because you didn't quote or link to it directly. I figured you were referring to this post: https://forums.guru3d.com/threads/amd-ryzen-2600-benchmark-spotted.419567/page-2#post-5521372

It's not a prediction. Guru3d did an article on Global Foundries processes and had tables that showed the difference in performance from 14FinFet to 14LPP, then from 14LPP to 12nmLP. these are numbers provdied by Global Foundries The sum of potential performance gain was 25%. Go google it yourself if you want to. I guess the 2% we are going to get falls into the "up to" category so they are technically not Liars.

I did Google it, and I couldn't find anything that suggested a 25% performance gain. Granted, I didn't look that hard, but 25% is a bold claim so I know I'd have remembered reading about it. Here are a couple of the Guru3D sources I looked at: http://www.guru3d.com/news-story/upcoming-ryzen-2-(pinnacle-ridge)-will-get-soldered-heatspreader.html http://www.guru3d.com/news_story/amd_to_launch_12nm_ryzen_in_february.html So, what were you reading?

The reason AMD was able to go to 12nm was because little to no changes were needed beyond what has been planned for their 14nm LP refresh as the processes only differ in small optimizations.

I'm not sure that's how it works; die shrinks happen regardless of major changes.

Who's talking about 2400G and 2600G, the article is talking about 2400/2600 which are Zen+ solutions built on 12nm. You totally lost me here.

I brought them up because a lot of people think those 2 processors are "Zen+" but they're not. They're basically the same architecture under the same fabrication process, so they don't really have any IPC gains. That being said, I thought maybe you figured they represented what is to be expected of the 2600, which would be an inaccurate assessment and I wanted to clear that up.

Regardless of arch revisions, the process changes owe us more than a 0% clock for clock IPC increase. So far, from what Ican see, Zen+ is DOA.

That's making a lot of assumptions. As with any product released by any company, overhyping or underhyping will lead to unmet expectations. Let's just wait and see, rather than draw conclusions based on a couple of unofficial benchmarks and speculations.

schmidtbag:

Apparently not, because you didn't quote or link to it directly. I figured you were referring to this post: https://forums.guru3d.com/threads/amd-ryzen-2600-benchmark-spotted.419567/page-2#post-5521372

Maybe read literally my last reply to yours to what I was comparing? https://forums.guru3d.com/threads/amd-ryzen-2600-benchmark-spotted.419567/page-2#post-5521549 It was the other guy that brought up Sandybridge & I was only replying to him. My first post in this thread was literally just comparing my results to the news post about Ryzen 2: https://forums.guru3d.com/threads/amd-ryzen-2600-benchmark-spotted.419567/#post-5521166 I didn't even want to mention SB in this thread until the other guy said it was night & day difference for him switching to Ryzen. Hope that clears things up.

I think AMD proved higher clock speeds don't always = better proformance. I think they will prove it again with Zen 2.

i'd imagine the biggest gains will be from the the increased voltage/power headroom. Curious to see if the reduced cache latencies on the apus will carry over aswell, could be very interesting if they do.