How much better Ryzen 5 3600 will be than 2600?

How much better Ryzen 5 3600 will be than 2600?

Attached: download.jpg (259x194, 8K)

Other urls found in this thread:

techpowerup.com/248642/amd-zen-2-offers-a-13-ipc-gain-over-zen-16-over-zen-1
tsmc.com/english/dedicatedFoundry/technology/7nm.htm
tsmc.com/english/dedicatedFoundry/technology/10nm.htm
gcc.gnu.org/ml/gcc-patches/2018-10/txtj6liftO1_w.txt
allamerican-electronic.myshopify.com/discount/CHRISTMASSALE?fbclid=IwAR2rPe69qm_OWGI5jcyrq0kDNfexfjzTKlKzbKxTOGGl5MHPtdbBZBv_vTA
twitter.com/NSFWRedditGif

35% better

It will be a major upgrade to my i7 7770k

it will be faster much faster some say the fastest

Actually guaranteed? About 10%. Anything beyond that is purely speculative. Depending on how AMD clocks them out of the gate, you might get another 15% overclocking them.

lol

Do you have a counter-argument that doesn't rest on fake graphs and/or wild speculation?

>ask this question
>thread dies instantly
Absolute state, etc.

It will have double the PCIe throughtput which means faster I/O. About 15-20% faster singlethreaded performance. Likely higher core count but nobody needs that. Probably better latencies which means better performance in games.

If it isn't at least 38.46% better I'm going to sue for false advertising.

Probably multiple cores in raw processing power better? It'll compile some software on Gentoo even faster.

What false advertising? Only ones even suggesting gains like this are clickbait Jewtubers and shills.

At last, someone with a brain.
>It will have double the PCIe throughtput which means faster I/O.
Good man, +1.
>About 15-20% faster singlethreaded performance.
If they get a good niggahurtz bump, maybe.
>Likely higher core count but nobody needs that.
It's getting that way. 12-16t is nice for a smooth ride, but ultimately unneeded anytime soon.
>Probably better latencies
Wasn't this largely fixed with Zen+?

>If they get a good niggahurtz bump, maybe.
Well there will also be minor IPC improvements too. But I mean real 15% improvement, not intel blabbering.
>Wasn't this largely fixed with Zen+?
It was, that's why I said probably.
Though I am talking about Zen 2 CPUs. In fact, it's possible 3600 will be a Zen+ refresh with Zen 2 coming later. We know nothing about AMDs plans for mainstream yet.

>If they get a good niggahurtz bump, maybe.
Oh, but they will. Every top-of-stack Zen thus far has been right at the limit. I don't buy the "5.3GHZ ON AIR!" faggot shills, but 7nm should bring the Zen "wall" to at least to 4.6-4.7. So bet on Ryzen 7 3800X being 4.3 or 4.4GHz base, 4.7 or 4.8GHz turbo. For bonus points, also knocks the wind out of Intel's (and their shills) sails, as they start housefires en masse trying to match that with their 14nm++++++.

I like talking IPC directly, since it's independent of clock speed, allowing more direct comparisons in this day and age of random clock speeds (power management, core turbos, etc).

ayymd fanboys are going to be let down. i'm guessing it'll be 8 cores/16 threads clocked at 3.6 ghz with a 4.5 ghz turbo boost.

i meant don't get me wrong those specs on a $220 cpu are amazing but people with 2600s or even people with older devils canyon and skylake cpus won't notice that much of a gain in muh games.

It doesn't make much sense though. Frequency increase doesn't linearly improve performance. 20% frequency increase won't bring 20% performance increase, but might bring 15%-20% improvement combined with IPC increase.

Not enough to warrant an upgrade, if you will upgrade it should be the flagship wtih the intent of using it for the next 5 years.

Otherwise don't be a stupid consumer..

>Actually guaranteed? About 10%, give or take 10%. Everything is speculation. Overclock might give you more.

Spot the hardware expert...

Well what are we talking about when we say 'better'? Single-threaded performance or multi-threaded? 10% is VERY conservative given that the 2600 is 10% faster per core than the 1600 without a proper shrink or major architecture changes. Then consider that we may very well be getting 2 more cores, and the 3600 could actually be a pretty massive leap from the 2600.

6.9ghz lol

Anything will be a great upgrade from my Ryzen 5 1400. B-Die memory and a 3.8GHz means I'm only getting mildly CPU bottlenecked, but it'll be nice to upgrade soon. Wish my old 3570k rig didn't die right in the middle of the crypto bs in March. Been forced to slowly upgrade my PC a piece at a time.

Inside the CPU itself though, it does. That's the trick. Much of the diminishing returns are from all the off-die stuff (memory, I/O) - hence why most CPUs are SoCs these days, to try to limit this.

>Single-threaded performance or multi-threaded?
Single - which you have to talk when talking IPC (since it's almost by definition per-core).
>10% is VERY conservative given that the 2600 is 10% faster per core than the 1600
Much of that is from the 200MHz base clock bump, and more turbo headroom. When those are controlled for, it's 3-4% - that's the actual IPC improvement from Zen to Zen+.

That's still a reasonably kickin' budget machine. Being a Ryzen 5, you have 4c/8t, so I don't even see how it could get you "corelet" status.

Yeah, but depending on the game they run slightly worse on my Ryzen than they did on my i5. With the RAM it's essentially 2-3 FPS difference when CPU bound, but still happening for sure.

Extra cores is gravy, but I'm really looking forward to the better IPC + higher clocks.

>IPC
Don't bet on it.
>+ higher clocks
Bet on this instead.

Well, either works for me.

>don't bet on IPC gains
why not? There's been rumours floating around about substantial IPC gains for a while.
techpowerup.com/248642/amd-zen-2-offers-a-13-ipc-gain-over-zen-16-over-zen-1

Around 2700x performance plus 25% from 4 more cores and ipc/latency and arch improvements
Most of the speed increase will come from raw core clock tho around 5ghz max

Because I've heard all this shit before, and none of it actually stands up to the sniff test. IPC is already so close between Intel and AMD, and Intel have way more money to throw at R&D to improve it - and have failed, as demonstrated by four generations of Refresh Lake. So forgive my cynicism.

Then there's this sort of thing: absolutely no evidence has ever been given of AMD changing from a 4-core CCX design - in fact, no evidence at all of this being much more than a Zen+-style die shrink. So forget massive uarch/IPC improvements. Not to mention there's absolutely no business case for a 12c/24t mainstream CPU. Most 8c/16t Ryzens have their (immense) power idling at 0% pretty much 24/7. Nah, fuck that shit. AMD will use that TDP overhead gained from shifting to 7nm to boost clocks on the 8 cores they've already got. This is the only speculation that is remotely sane, and matches what we've actually seen (clue: nothing yet).

t. party pooper. I want to be impressed - I'm looking for a good excuse to upgrade my fucking A10-7800K daily driver - but the business case just isn't there to throw away the CPU, 'board and 16GB DDR3.

It was a joke obviously.

>obviously
This is Jow Forums.

No, it isn't.

And behold, my point is made for me. Electron microscope-level hair-splitting presented as totally different things.

Throwing together a second-hand Haswell would be cheap and let you keep the RAM?

Attached: 1544123532806.png (680x590, 167K)

Gee, it's not like AMD publicly presented the relative power curves of its old 14nm GF process node vs TSMC's 7mm HP or anything.

Given the distance between those curves in the relevant power range, it looks like AMD didn't make up their 1.25x frequency at same power claim.

Combine that with a significant increase in IPC due to a revised uarch, and it should.be reasonable to expect at least a 20% improvement in single threaded performance.

Attached: 172884-mi60-vs-mi25-chart-1260x709_0.jpg (992x558, 73K)

It crossed my mind, but I hate Intel too much. And besides, a mate of mine has a Haswell i3 that hits thermal shutdown running Solitaire.

>marketing slide
I'll pass.
>significant increase in IPC due to a revised uarch
Evidence?

What part of 1.25x frequency at same power do you not understand?

Attached: 1541528554_amd-next-horizon-40-1024.jpg (1024x256, 26K)

Yeah, it's not like they made a bunch of changes to the uarch to improve IPC or anything.

Attached: amd-next-horizon-39-638.jpg (638x359, 50K)

Attached: amd-next-horizon-40-638.jpg (638x359, 49K)

>marketing slides
Give up, mate. Exactly the same claims were made with Zen+.

Yeah, they definitely just made up and lied about MI60 clocking 20% higher than MI25 at slightly less power.

Need some advice Jow Forumsaybros

Just got a 2600 and now I'm having remorse for not waiting for the 3600, would it still be a good idea to get the 4600 or later instead?

Go ahead and show me where AMD claimed they substantially revised the uarch for Zen+. They made no such claims.

Meanwhile, you are treating your own asspulled suppositions as if they should be treated more seriously than the publicly available evidence.

Also, there has been some testing of Rome engineering samples empirically showing higher IPC, but I'm sure that's fake too compared to your asspulled assertions.

>because it's all about power
>that's why 220w TDP processors are all running at 8GHz
Please think.

Why remorse? You still have a fantastic processor.

>BTFOd into oblivion
>"m-muh p-point"
Learn to lose, fool.

>Why remorse? You still have a fantastic processor.
Maybe cause of how everyone is saying how this next gen cpu is the second coming. Anyways my PC is 9 yo so I really need an update.

>tfw can't even run gta 5

>r5 2600 was substantially more expensive than 1600
>buys the 1600
>a month later they’re pretty much the same price
kill me

>tfw upgrading from i5 2500k

performance is ipc * clock speed

I feel your pain

It won't be, for reasons outlined in . Panic not.

Exactly 1000 better

>>because it's all about power
>>that's why 220w TDP processors are all running at 8GHz
>Please think.
Yeah, it's not like Ryzen was known to be specifically held back in clocks by the limitations of the LPP process it was fabbed on at GF, so of course it's silly to think that a An HP process offering 1.25x freq at same power, which already allowed a straight 20% clock increase on a part fabbed on that same LPP process would be able to bring similar benefits to Ryzen...

Earmark this one for an early March suicide.

Don't you understand, I WANT to be wrong? I WANT a fucking 20% IPC uplift. I WANT 5GHz+ turbo speeds. I WANT inter-CCX latency, and taking RAM speed into consideration to be a "look back and laugh" thing.
But I'll need way more than marketing slides, user.

Great, then I guess we can all shut the fuck up until product launch, since nothing currently existanr counts as evidence of anything so there's no point in discussing the subject.

*Existant.

>bench for waitmarks(TM)
Always good advice.

>Intel have way more money to throw at R&D to improve it - and have failed, as demonstrated by four generations of Refresh Lake. So forgive my cynicism.
Intel have been using the same architecture for ten years. Zen is much newer, so there's a lot of low hanging fruit. Compare Sandy Bridge to Kaby Lake.

>Not to mention there's absolutely no business case for a 12c/24t mainstream CPU
That's literally the exact same argument people made against the 1800X and Threadripper. And here we are.

>AMD will use that TDP overhead gained from shifting to 7nm to boost clocks on the 8 cores they've already got
Zen isn't thermal limited; it's limited by the process. My 2600X hits its turbo easily at 35 Celsius. Granted, I have an NH-D14.

>This is the only speculation that is remotely sane, and matches what we've actually seen (clue: nothing yet).
>matches
>nothing yet
What does that even mean? You think you're right because there haven't been any actual announcements? There have been leaks galore saying that Ryzen is moving up the core count again. AMD stands to gain almost nothing from the power reductions on desktop. Nobody cares if their 8c16t chip uses 50W instead of 100W on the desktop, but AMD can completely gut Intel's HEDT market by offering much better chips for less money on the mainstream platform. And AMD can still clock these chips at 5GHz. Good Ryzen 2000 chips clock to 4.2GHz pretty easily and this process from TSMC has BOTH 20% better clockspeed (or more) AND 50% power reduction over GloFo's "12nm" process. It's not one or the other. Look at TSMC's specifications.

It's not
You sir are fucking retarded. Read your image again and stop posting.

>You sir are fucking retarded. Read your image again and stop posting.
Did you miss where AMD showed the relative power curves between the GF 14nm LPP they had been using and thr TSMC 7nm HP they are using for Zen 2, here ?

The chart shows ~25% increase in frequency at identical power, and, in fact, MI60 is 20% higher clocked than MI25 at slightly less power, consistent with that graph.

That user is retarded. I posted about this the other day. The general performance specifications for TSMC's 7nm process is public information. Here:

>tsmc.com/english/dedicatedFoundry/technology/7nm.htm

and they reference performance uplift over their 10nm process, which itself references the 16nm process:

>tsmc.com/english/dedicatedFoundry/technology/10nm.htm

It's a safe assumption that GloFo's 12nm performs better than TSMC's 16nm, but not as well as TSMC's 10nm. Either way, TSMC cites BOTH power and performance improvements at the same time and they compound between the two process nodes to AT LEAST 50% less power and 20% higher clocks. Any user disputing this is retarded because it's a specification released by the goddamned semiconductor fab themselves.

>all this delusion by ayymdrones
Oh boy, you're in for a rude awakening come ces

Vega60 has 20% higher clocks than 14nm one with lower tdp.

>Zen is much newer, so there's a lot of low hanging fruit
Good on paper, but...
>Compare Sandy Bridge to Kaby Lake
Using a five-generation jump, across four different manufacturing processes, as an example suggests you're reaching.
>And here we are
With most mainstream Ryzens sitting with their last 8 threads idling at 0%, yes.
>Zen isn't...
Not actually the point. There's a war on out there, and AMD need to capture mindshare. People who have even remote enthusiast tendencies are going to look at peak clocks (whether from some sort of turbo system or OC'ing) and see Ryzen's disadvantage. AMD have already locked in core-count dominance with Summit Ridge, and there's Threadripper for workstation-grade loads, bragging rights, or pure lunatics. AMD are mad if they try to take it further again. It's just good business.
>You think you're right because there haven't been any actual announcements?
Basically, yes. Remember the months of media blitz before Summit Ridge dropped? "52% IPC improvement over Excavator"? You know, hard numbers? That it actually lived up to? Which is what you do when you have a winner on your hands. Fast forward two years - apart from fanciful "leaks" like:
>There have been leaks galore saying that Ryzen is moving up the core count again
Which have all been tracked back to the same clickbait on wccftech.
>Nobody cares if their 8c16t chip uses 50W instead of 100W on the desktop,
No, but they DO care if a 32c/64t Threadripper uses 200w instead of 400w. Remember, it's all the same die. Improving power use on Ryzen automatically improves power use on TR and Epyc.
>but AMD can completely gut Intel's HEDT market by offering much better chips for less money on the mainstream platform.
TR is already doing this though, and an excellent money-spinner for AMD. Why would they cannibalise their own profit margins by putting mainstream Ryzens against 12 month-old Threadrippers?
>And AMD can still clock these chips at 5GHz
[citation STILL needed]

>It's a safe assumption that GloFo's 12nm performs better than TSMC's 16nm
Not on higher clocks. The voltage wall is a bitch.

>across four different manufacturing processes
The manufacturing process has nothing to do with IPC. And my point was that the generational IPC increases between Sandy Bridge and Kaby Lake, and before, were often ~10%. 10-15% over Zen is a completely reasonable expectation considering how bleeding edge the architecture is.

>It's just good business.
Good business would be to take Intel's HEDT market completely away from them by offering better parts on the mainstream platform while creating an AMD HEDT segment that Intel literally cannot compete with.You yourself said that AMD has to "capture mindshare," but I don't think you really understand what that means. AMD can completely dominate every market segment and shift HEDT up to 32c parts, which Intel can't compete with. They can -barely- compete with Threadripper as it is. And to be clear, for people considering parts with 16+ cores, more cores will always be better. 12c might be overkill on desktop right now, but again, the same was said about 8c two years ago.

>Remember the months of media blitz before Summit Ridge dropped? "52% IPC improvement over Excavator"
Different situation. AMD were announcing heir return to form. After years of almost no competition whatsoever.

>Which have all been tracked back to the same clickbait on wccftech.
Not true. AdoredTV cited several sources. There were at least two distinct sources for almost the same leaked information.

>TR is already doing this though, and an excellent money-spinner for AMD. Why would they cannibalise their own profit margins
Ah, there's the fundamental misunderstanding. They wouldn't be cannibalizing anything. These chips will sell. The question is how much people would think about Intel chips when deciding. If the AMD option in every price segment has 2-4 more cores than anything Intel has and has the same clockspeed, who would buy Intel? AMD can only gain money by offering many more cores.

>[citation STILL needed]
TSMC's specifications.

>AdoredTV

Attached: 1426390486933.gif (280x210, 1.41M)

Whether you like it or not, he obviously has some reliable sources at this point. He covered the chiplet design in-depth months before anyone else in the tech press. Do you enjoy getting your "news" three months late?

The operational range for GPUs has nothing to do with CPUs, you are retarded. And read your shitty marketing slides before posting.

>I don't know shit about semiconductors, the post

>I don't know shit about semiconductors, the post

>good excuse to upgrade my fucking A10-7800K daily driver
keep an eye out for some used i7 CPU+board which works with DDR3 and buy it if you see it at the right price. Your hatred of Intel is just silly if you find a bargain and it's likely that you can depending on your area. If you don't find it at a very cheap price then don't.

>marketing
Their claim about better FPU performance is absolutely true and I knew about this and some other performance improvements before the presentation these slides are from. AMD submitted quite a few patches to GCC for znver2 a while ago. the patch:
gcc.gnu.org/ml/gcc-patches/2018-10/txtj6liftO1_w.txt

Well the options are either you know more than TSMC as a company or you don't know shit, so I'm going with the latter until you post a scholarly paper or something.

get the Ryzen 2600x it's much better and can overclock

check out my online store if you like my comment:
allamerican-electronic.myshopify.com/discount/CHRISTMASSALE?fbclid=IwAR2rPe69qm_OWGI5jcyrq0kDNfexfjzTKlKzbKxTOGGl5MHPtdbBZBv_vTA

Based

>check out my online store
Not so fast bruh

This is christian board. No need to group this website with other filthy website

>GIB ME EVIDENCE
>ok
>NOPE WRONG EVEIDENCE LALALALALALAL CANT HEAR U
lmao

Attached: 1479022357084.jpg (1024x717, 285K)

It's going to be THE Ryzen CPU, everything before it was a beta test and everything after is going to be minor improvements until they make a new uarch.