AMD Computex 2019 Livestream

trentchau said:
I think it's the smarter move from AMD to release a 12 core imho. Just enough for what feels like an upgrade worth the 66% increase in price, but not enough to make people who buy your more mainstream $329 product think they are getting shafted. As much as power users feel like we matter to the bottom line, it's the $200 spenders that make companies money. If AMD immediately release a 16 core chip, it basically would tell their $329 users they are the new "Average". Could bite them in the butt.
Click to expand...

Yeah, it makes sense to leave the 12c as the highest end for now. Having the best is very important in these markets and if AMD's promises hold true the 12c R9 is going to be a monster at $500 and get AMD all the media and consumer attention. Then they'll have that 16c in their back pocket that they can use to get even more attention down the road.
 
next-Jin said:
It would be in an entirely different category though and wouldn’t be for gamers. The clocks would be lower I’d imagine.
Click to expand...

If the rurmors are true and Threadripper is removed, I think gamers would think Ryzen 9 is within their domain. Look at all the i7's from the last 10 years that gamers got, even though the i5's were probably just fine for them. I feel the best way to differentiate being a different catagory is a physical limitation where a chip isn't compatible and requires a different chipset.
 
The biggest news here is that AMD is beating Intel's best single threaded numbers. Only by a tiny margin, but still, that is huge. I'm buying one.

trentchau said:
using a none HT chip for comparison? Really.
Click to expand...
lightsout said:
I know is this a joke? Why even do it then?
Click to expand...

There is something to be said for comparing CPU's at the same price point.
 
AMD's Robert Hallock showing off Ryzen 9 3900X.

azLb1a4IrPQuyyNHaPm9RihDS5roZXfMIOTl8e5nKkA.jpg
 
trentchau said:
a little off topic, but is the heat really that bad on the i9-9900k? (Asking like a complete noob). I have a air cooler on my 5960x with a 4.5 all core oc and it really doesn't have much thermal issues. I don't own a 1151 chip that does that much tdp though, but I've heard people talking about the heat issue like it's a problem.
Click to expand...

Generally speaking no- one thing to consider is the customer base: those spending on a 9900K or a twelve- or sixteen-core Ryzen don't just want it clocked high and running cool, they want it done quietly, speaking from experience. That means overdoing it a little bit. A top-shelf air cooler (still <US$100) and a case with decent front to back airflow would be enough to hit boost speeds out of the box.
 
Zarathustra[H] said:
The biggest news here is that AMD is beating Intel's best single threaded numbers. Only by a tiny margin, but still, that is huge. I'm buying one.




There is something to be said for comparing CPU's at the same price point.
Click to expand...
But if they really are beating IPC. They could have showed that against the 9900k. Odd choice. I get the price thing though.
 
lightsout said:
But if they really are beating IPC. They could have showed that against the 9900k. Odd choice. I get the price thing though.
Click to expand...

IPC is one thing- they still can't clock high enough, so they'd still be behind on average for single-threaded workloads.

Not that they'd be so far behind that it matters for more than benchmarks (finally!), but this is all marketing we're talking about right now.
 
Ryzen 9 3900X - 12C/24T, 105W TDP, 3.8 GHz Base, 4.6 GHz Boost - 6 MB L2 + 64MB L3

We got the chip, now all we need are the benchmarks. AMD is looking good.
 
  • Like
Reactions: blkt
like this
I'm excited about this release! The Ryzen 9 should almost match the performance of the 2950X!

The reason Threadripper 3000-series has been delayed is because AMD already played this trick they're using on Ryzen 9 and EYPC 64-core, and doubled the chip count. They can't realistically double it again (and still have enough memory bandwidth, and reasonable e prices)

So Threadripper 3 is looking to be the same 32-core setup, but with a lot better multi-core scaling (single unified memory controller, and better interconnect). The most credible rumor puts the delay until 2020 so they can launch the platform on Zen 2+, and maximize per-core performance. It also means they can have the "optimized" yields, the same thing that allowed them to release Threadripper 2000-series at a reasonable price.
 
Last edited:
GreenOrbs said:
I'm very happy that AMD has finally caught up in ST performance. If reviewers show that the new Ryzen 7 series matches or beats out Intel 9900k single threaded performance by 1-3% as AMD claims there's really little reason to buy Intel. Sure the crazy rumors of 5GHz speeds didn't come to pass but it seems like the 15% IPC gain made up for it.
Click to expand...

Well its a tradeoff. +15% IPC but lower clocks. I mean if anything it is catching up, but the price point looks great!
 
  • Like
Reactions: blkt
like this
IdiotInCharge said:
IPC is one thing- they still can't clock high enough, so they'd still be behind on average for single-threaded workloads.

Not that they'd be so far behind that it matters for more than benchmarks (finally!), but this is all marketing we're talking about right now.
Click to expand...

I think the only thing that bothered me about it is they went for the "Look how much faster we are option". If I saw that their $329 cpu is lightly faster than Intels $500 cpu, or equal to it...that would hold more value to me imho. I can't help but feel like AMD was a little more Shady in their competitor choice in that test.

Added to the fact that dude was like "See our competitor hasn't even finished yet" before he left the stage. It felt kinda childish.
 
defaultluser said:
I'm excited about this release! The Ryzen 9 should almost mach the performance of the
2950X!

The reason Threadripper 3000-series has been delayed ois because AMD already played this trick they're using, and doubled the core count. They can't realistically double it again (and still have enough memory bandwidth)

So Threadripper 3 is looking to be the same 32-core setup, but with a lot better multi-core scaling (single unified memory controller, and better interconnect). The most credible rumor puts the delay until 2020 so they can launch the platform on Zen 2+, and maximize per-core performance.
Click to expand...
Sounds reasonable. 7nm+ from TSMC would be great for it. :)
 
3700X to 3800X
3.6 GHz base to 3.9 GHz base (+300 Mhz) and 4.4 GHz to 4.5 GHz (+100 Mhz) for 61.5% more power (65W to 105W). Yikes. I think OCing these will be like existing Zen+ chips: Don't bother, let the PBO algorithm take care of it.
 
trentchau said:
I think the only thing that bothered me about it is they went for the "Look how much faster we are option". If I saw that their $329 cpu is lightly faster than Intels $500 cpu, or equal to it...that would hold more value to me imho. I can't help but feel like AMD was a little more Shady in their competitor choice in that test.

Added to the fact that dude was like "See our competitor hasn't even finished yet" before he left the stage. It felt kinda childish.
Click to expand...

I'd recommend getting used to it- not because AMD does that (they do), but because this is a marketing fight, so everyone does it because they cannot afford not to. It's like politics, but with more money.

And there's a pretty thin line between those that don't know enough to know, and those that know not to take marketing slides to heart.

Based on what AMD released, we can reasonably surmise that they didn't pull off a performance miracle- but really, delivering a competitive product is winning for them. And hell, they've shown that generally speaking, they have an inexpensive means to produce CPUs with more cores than has been the norm. It's going to cost Intel to compete on core counts.
 
Low TDPs are a good sign and released numbers right within what I had predicted. Can't wait to see OC..
 
I wonder why AMD didn't just use single socket epyc in place of threadripper? Just yank the BMC features and any other crap HEDT users don't need?
 
IdiotInCharge said:
IPC is one thing- they still can't clock high enough, so they'd still be behind on average for single-threaded workloads.

Not that they'd be so far behind that it matters for more than benchmarks (finally!), but this is all marketing we're talking about right now.
Click to expand...

They claim IPC is up 15% over previous Ryzen.

They just claimed a 1% advantage for the Ryzen 7 3700x vs Core i5-9700k in single threaded Cinebench,

That's pretty amazing considering they claim the 3700x has a max turbo clock of 4.4Ghz, and from spec sheets and the i5-9700k has a max turbo clock of 4.9Ghz.

If they are 1% faster at 4.4Ghz than Intel is at 4.9Ghz, if my post-vodka calculations are correct, it means they are beating Intel by ~12.5% in IPC.

Is this even possible?
 
Zarathustra[H] said:
Is this even possible?
Click to expand...

Possible? We should be well past 10% IPC over Skylake by now, just in general. AMD appears to have whacked their teething problems with Ryzen (statement pending availability...), but at the same time, they compared a sixteen thread Ryzen to an eight thread Coffee Lake (aka Skylake 3).

So they may be faster at or near the same price, but they may also not be as fast as they're loosely implying.
 
Zarathustra[H] said:
They claim IPC is up 15% over previous Ryzen.

They just claimed a 1% advantage for the Ryzen 7 3700x vs Core i5-9700k in single threaded Cinebench,

That's pretty amazing considering they claim the 3700x has a max turbo clock of 4.4Ghz, and from spec sheets and the i5-9700k has a max turbo clock of 4.9Ghz.

If they are 1% faster at 4.4Ghz than Intel is at 4.9Ghz, if my post-vodka calculations are correct, it means they are beating Intel by ~12.5% in IPC.

Is this even possible?
Click to expand...
Where did you see single threaded cinibench result?
 
lightsout said:
Where did you see single threaded cinibench result?
Click to expand...

From the AMD slide deck

COMPUTEX_KEYNOTE_DRAFT_FOR_PREBRIEF.26.05.19-page-035b_575px.jpg


We will have to wait until July to verify them in real reviews, but it was exciting to watch live!

Also, Intel' s Emergency Edition 9900KS is going to close the gap, but not exceed the Ryzen 9's sizeable lead. I see a price cut from them in the near future.

AMD can just save the 16-core for next year's Zen 7nm+ refresh, where it can counter Comet Lake..
 
Last edited:
Dayaks said:
I think I am too lazy to pull my 2700x for a 12C part. 16C probably would.

I hope these do as well as AMD is saying. Time will tell.
Click to expand...


IMHO, the core count is kind of the boring news here. The +15% IPC while at the same time improving the max turbo clock are what I am drooling over.
 
IdiotInCharge said:
Possible? We should be well past 10% IPC over Skylake by now, just in general. AMD appears to have whacked their teething problems with Ryzen (statement pending availability...), but at the same time, they compared a sixteen thread Ryzen to an eight thread Coffee Lake (aka Skylake 3).

So they may be faster at or near the same price, but they may also not be as fast as they're loosely implying.
Click to expand...


Well, what I meant was, they claim to have improved IPC 15% over Zen+, and the numbers they are claiming put them at a 12.5% IPC advantage over Skylake.

Were they really that close to Skylake before?

The numbers don't make sense to me.
 
odditory said:
Don't get your hopes up. If their GPUs are anything to go by, it'll be closer to maxed out from the factory than not.
Click to expand...

True. These things only get more and more difficult as the dies shrink. If 7nm holds up the trend, pobably not much. At least not at first. Maybe down the road yields will improve enough to allow later buyers better overclocks.
 
You must log in or register to reply here.
Back
Top