Jump to content

(16core added)AMD 3000 specs! 4.7 GHZ, R9 3950x, R7 3700x, 3800x.

13 hours ago, ThePD said:

The leaks were completely unrealistic. I am happy for such high IPC improvement with the massive drop in power consumption. That 3900X tho... Goddamn

 

i still dont get how were the rumors unrealistic, intel has 5GHz chips why would it be impossible to think amd could have one too?

new node, new architecture, why is it so unbelievable?

 

obviously it didnt happen, but to say it was unrealistic is kinda stupid

MSI GX660 + i7 920XM @ 2.8GHz + GTX 970M + Samsung SSD 830 256GB

Link to comment
Share on other sites

Link to post
Share on other sites

4 minutes ago, Neftex said:

i still dont get how were the rumors unrealistic, intel has 5GHz chips why would it be impossible to think amd could have one too?

new node, new architecture, why is it so unbelievable?

 

obviously it didnt happen, but to say it was unrealistic is kinda stupid

Because Intel has had 9 iterations of optimization to get to 5GHz, Intel is also known as the best fab technology leader and has their nodes specifically designed for their arch and high clocks. Architecture also matters a great deal for maximum clocks too, Zen 2 is still only a revision of Zen so expecting a 1Ghz jump from a single generation is a bit much, nodes help a lot with things like power and density but they don't directly translate to higher clocks. 

Link to comment
Share on other sites

Link to post
Share on other sites

2 minutes ago, leadeater said:

Because Intel has had 9 iterations of optimization to get to 5GHz, Intel is also known as the best fab technology leader and has their nodes specifically designed for their arch and high clocks. Architecture also matters a great deal for maximum clocks too, Zen 2 is still only a revision of Zen so expecting a 1Ghz jump from a single generation is a bit much, nodes help a lot with things like power and density but they don't directly translate to high clocks. 

it was new node we havent seen yet, so we had nothing to judge from, and pretty much noone here knows the changes to zen2 architecture to say it was outside of realms of possibility

MSI GX660 + i7 920XM @ 2.8GHz + GTX 970M + Samsung SSD 830 256GB

Link to comment
Share on other sites

Link to post
Share on other sites

23 minutes ago, leadeater said:

Games that should be the ones using the most cores possible, bloody Firaxis.

Those just dislike my TRs big time.  Aka, those will crash if I don't set core affinity and restrain to one die.

Reason I am curious about this 3900X and its internal layout.  Be interesting to see how those handle memory access and latency issues.

 

23 minutes ago, VegetableStu said:

Zen 2 shouldn't be seeing severe NUMA setbacks like those from 2990WX I think o_o there's a single die to interface all the memory channels and distribute to the CPU cores (rather than having the memory controller on the core dies themselves)

I'm not referring to 2990WX (those are interesting, but I would never pay the price for such a severely neutered Eypc).  I have a 1900X and 1950X and certain games straight up dislike running on them if I don't tune core affinity and memory access settings.  I do hope that IO die does help these new chips.  If it does, AMD better get a new TR out with it (plus, put it on the 7nm+ EUV as well please).

2023 BOINC Pentathlon Event

F@H & BOINC Installation on Linux Guide

My CPU Army: 5800X, E5-2670V3, 1950X, 5960X J Batch, 10750H *lappy

My GPU Army:3080Ti, 960 FTW @ 1551MHz, RTX 2070 Max-Q *lappy

My Console Brigade: Gamecube, Wii, Wii U, Switch, PS2 Fatty, Xbox One S, Xbox One X

My Tablet Squad: iPad Air 5th Gen, Samsung Tab S, Nexus 7 (1st gen)

3D Printer Unit: Prusa MK3S, Prusa Mini, EPAX E10

VR Headset: Quest 2

 

Hardware lost to Kevdog's Law of Folding

OG Titan, 5960X, ThermalTake BlackWidow 850 Watt PSU

Link to comment
Share on other sites

Link to post
Share on other sites

 

8 hours ago, Humbug said:

AMD's Gaming benchmarks

 

R7 3800x vs R7 2700x

Below results are from the combination of 200Mhz clockspeed bump and IPC improvements.

  • 34% uplift in CS:GO
  • 33% uplift in League of Legends
  • 31% uplift PUBG
  • 21% uplift in Overwatch
  • 15% uplift in DOTA2
  • 11% uplift in GTA5

 

 Show me a FFXIV v5 benchmark https://na.finalfantasyxiv.com/benchmark/ . MMO games trade off visuals for frame rate when actually online.

 

 

 

Link to comment
Share on other sites

Link to post
Share on other sites

29 minutes ago, Neftex said:

it was new node we havent seen yet, so we had nothing to judge from, and pretty much noone here knows the changes to zen2 architecture to say it was outside of realms of possibility

We had TSMC information regarding 7nm which is where people incorrectly took that information and tired to use the quoted increased for the clock increase. We've also never had an architecture revision give 1Ghz increase to product specs before.

 

Intel 32nm to 22nm didn't have base or boost clock increase at all. Sandy Bridge 2700K was 3.5/3.9 (32nm), Ivy Bridge 3770K was 3.5/3.9 (22nm), Haswell 4770K was 3.5/3.9 (22nm), Haswell refresh was 4.0/4.4 (22nm). Nodes simply don't give increased clocks like people want them to, Skylake (14nm) even managed to regress on Haswell refresh to 4.0/4.2.

 

Edit: Fixed Haswell base/boost

Link to comment
Share on other sites

Link to post
Share on other sites

that R5 3600X boost seems kinda low, Given its a 6 core and so it should have more headroom in TDP.

Good luck, Have fun, Build PC, and have a last gen console for use once a year. I should answer most of the time between 9 to 3 PST

NightHawk 3.0: R7 5700x @, B550A vision D, H105, 2x32gb Oloy 3600, Sapphire RX 6700XT  Nitro+, Corsair RM750X, 500 gb 850 evo, 2tb rocket and 5tb Toshiba x300, 2x 6TB WD Black W10 all in a 750D airflow.
GF PC: (nighthawk 2.0): R7 2700x, B450m vision D, 4x8gb Geli 2933, Strix GTX970, CX650M RGB, Obsidian 350D

Skunkworks: R5 3500U, 16gb, 500gb Adata XPG 6000 lite, Vega 8. HP probook G455R G6 Ubuntu 20. LTS

Condor (MC server): 6600K, z170m plus, 16gb corsair vengeance LPX, samsung 750 evo, EVGA BR 450.

Spirt  (NAS) ASUS Z9PR-D12, 2x E5 2620V2, 8x4gb, 24 3tb HDD. F80 800gb cache, trueNAS, 2x12disk raid Z3 stripped

PSU Tier List      Motherboard Tier List     SSD Tier List     How to get PC parts cheap    HP probook 445R G6 review

 

"Stupidity is like trying to find a limit of a constant. You are never truly smart in something, just less stupid."

Camera Gear: X-S10, 16-80 F4, 60D, 24-105 F4, 50mm F1.4, Helios44-m, 2 Cos-11D lavs

Link to comment
Share on other sites

Link to post
Share on other sites

2 hours ago, descendency said:

Clearly a 16 core model is possible, so it begs the question why not announce it today? Was it going to be $500 and AMD changed their minds due to Intel's recent hardships? 

 

Early July will be great. I can't wait to see how these chips perform. 

thats at least part of it, another part is to allow them to extend the coverage window, that way they can make a fuss about it later on, another thing is that this way they can keep tdps low, this should allow ryzen 3000 to work well on older boards, and maybe allow them to release the 16 core only for x570 with the excuse of being an extreme edition

2 hours ago, Drak3 said:

Maybe from the chiplet configuration aspect, but it might not be from the aspect of thermals or power draw, the socket itself might not be able to handle it. And that's something we won't know until either overclock push the socket itself to its limit or AMD says something.

the socket should not be the problem, as many sources already confirmed the 16 core is coming they just dont know when, its probably more like a problem with supporting x370 and b350 which have a good change of going boom if you use the 16 core on them and oc it

1 hour ago, leadeater said:

They don't actually need all that great dies, server clocks are quite a bit lower than desktop so you don't actually need clock binned dies.

the server still expects top silicon, as it still helps to lower consumption and increase longevity, and they do pay for it

45 minutes ago, Neftex said:

it was new node we haven't seen yet, so we had nothing to judge from, and pretty much noone here knows the changes to zen2 architecture to say it was outside of realms of possibility

we do have radeon VII which got a good frequency boost

Link to comment
Share on other sites

Link to post
Share on other sites

Are the boost clocks for all cores or single-core boost? I seem to remember hearing it was all core boost but I haven't been able to find a source to confirm it.

Link to comment
Share on other sites

Link to post
Share on other sites

1 minute ago, inquam said:

Are the boost clocks for all cores or single-core boost? I seem to remember hearing it was all core boost but I haven't been able to find a source to confirm it.

expect it being single core boosts

Link to comment
Share on other sites

Link to post
Share on other sites

1 minute ago, cj09beira said:

 expect it being single core boosts

Kinda cool that it keeps pace with the 9900K then that does a four-core 4.7GHz boost. Would indicate the architectural changes matters quite a lot.

Link to comment
Share on other sites

Link to post
Share on other sites

2 hours ago, MeatFeastMan said:

I think it's not what we think it is. People are saying they are holding back the 16 core in case Intel have a good answer. I think there's a bit of that going on, but..I think that if they released the 16 core now, the 12 core sales would suffer. It's kind of like threadripper. Who buys the 2970WX? Nobody. They either buy the 2950x or the 2990WX.

 

And it's the same situation here. They'll want to maximize 12 core sales by making it the temporary flagship, and then once the sales are where they want them, they'll add on the 16 core. Otherwise everybody would buy the 16 core or the 8 core, and forget the 12 core.

 

 

And they want to wait a bit for slightly better yields as well most probably as well

Link to comment
Share on other sites

Link to post
Share on other sites

49 minutes ago, leadeater said:

We had TSMC information regarding 7nm which is where people incorrectly took that information and tired to use the quoted increased for the clock increase. We've also never had an architecture revision give 1Ghz increase to product specs before.

 

Intel 32nm to 22nm didn't have base or boost clock increase at all. Sandy Bridge 2700K was 3.5/3.9 (32nm), Ivy Bridge 3770K was 3.5/3.9 (22nm), Haswell 4770K was 3.5/3.9 (22nm), Haswell refresh was 4.0/4.4 (22nm). Nodes simply don't give increased clocks like people want them to, Skylake (14nm) even managed to regress on Haswell refresh to 4.0/4.2.

 

Edit: Fixed Haswell base/boost

1GHz could have been possible here since its not just a node change, but also a manufacturer change. TSMC seemed to have better clocks (at least on GPUs) than GloFo even on the last node, so that would've been a 500MHz-700MHz increase instead.

That being said... I don't understand why they just don't seem to be in a hurry to launch their products these days... especially with navi. We all know it wont be groundbreaking, just release it to be okay, and move on to the next gen which is where we can start to have bigger hopes and dreams!

Link to comment
Share on other sites

Link to post
Share on other sites

Some of the expectations were so ridiculous.  People must have thought AMD had some physics defying magic up their sleave. 

i9-9900k @ 5.1GHz || EVGA 3080 ti FTW3 EK Cooled || EVGA z390 Dark || G.Skill TridentZ 32gb 4000MHz C16

 970 Pro 1tb || 860 Evo 2tb || BeQuiet Dark Base Pro 900 || EVGA P2 1200w || AOC Agon AG352UCG

Cooled by: Heatkiller || Hardware Labs || Bitspower || Noctua || EKWB

Link to comment
Share on other sites

Link to post
Share on other sites

10 minutes ago, TahoeDust said:

Some of the expectations were so ridiculous.  People must have thought AMD had some physics defying magic up their sleave. 

I have it good on industry insiders that the IPC increase is so yuge that a single core 1GHz Ryzen 3 outpaces a 5GHz 9900K, but it clocks over 5 petahertz.

Come Bloody Angel

Break off your chains

And look what I've found in the dirt.

 

Pale battered body

Seems she was struggling

Something is wrong with this world.

 

Fierce Bloody Angel

The blood is on your hands

Why did you come to this world?

 

Everybody turns to dust.

 

Everybody turns to dust.

 

The blood is on your hands.

 

The blood is on your hands!

 

Pyo.

Link to comment
Share on other sites

Link to post
Share on other sites

Interesting, but I'll wait until the reviews come before I decide that I can't afford an upgrade.

Link to comment
Share on other sites

Link to post
Share on other sites

2 hours ago, cj09beira said:

the server still expects top silicon, as it still helps to lower consumption and increase longevity, and they do pay for it

When a die isn't being pushed beyond what it can handle for clocks the difference between a good one and a bad one in terms of which can reach a higher clock isn't that different.

Link to comment
Share on other sites

Link to post
Share on other sites

3 hours ago, laminutederire said:

. I don't understand why they just don't seem to be in a hurry to launch their products these days... especially with navi. We all know it wont be groundbreaking, just release it to be okay, and move on to the next gen which is where we can start to have bigger hopes and dreams!

It's not good business to do that if you don't have anything better.  If AMD release another O.K gpu into the consumer market, then they are basically cementing that it's not a performance oriented product.  If they can't produce a decent GPU next then less people will wait to see what they release after that, they'll just buy Nvidia and be done with it.  We're already seeing posts of people saying they are tired of the "AMD hype" that doesn't bear fruit and waiting for less than desirable products,  not to mention the countless posts of people claiming they are the budget alternative.  Reputations of shitty products are very hard to shake,  we still have people saying don't buy AMD because slow and hot with dodgy drivers (that was like 10 years ago (pulling a number from my arse to highlight the issue)).

 

 

 

 

Nvidia can do it because the last 1.5 years of products are all still at the top of the performance graphs. So the market knows anything with 50 or below in the name is supposed to perform worse even if it isn't cheaper.

Grammar and spelling is not indicative of intelligence/knowledge.  Not having the same opinion does not always mean lack of understanding.  

Link to comment
Share on other sites

Link to post
Share on other sites

8 hours ago, Connor Price said:

 

If gaming performance is even similar to what I have now with my 8600k, I'm jumping ship. The 8600k isn't bad by any stretch but honestly after all of this spectre and meltdown bollocks I've definitely felt a drop in performance. It's even starting to chug a little with a bunch of tabs open and then switching to something like steam. Fans will ramp up and there's a little pause before anything happens

 

Placebo.. or something else

 

Spectre and Meltdown were never a big hit on Coffee Lake CPU's, maybe a couple of percents at most, but now with Windows 10 May Update it's completely mitigated and performance is back to normal

 

And you're not impacted at all with Zombieload and all the MDS stuff because your CPU doesn't even have hyperthreading

 

Not saying you shouldn't upgrade, that's up to you, but don't fool yourself

Quote or Tag people so they know that you've replied.

Link to comment
Share on other sites

Link to post
Share on other sites

5 hours ago, Ithanul said:

Those just dislike my TRs big time.  Aka, those will crash if I don't set core affinity and restrain to one die.

Reason I am curious about this 3900X and its internal layout.  Be interesting to see how those handle memory access and latency issues.

 

I'm not referring to 2990WX (those are interesting, but I would never pay the price for such a severely neutered Eypc).  I have a 1900X and 1950X and certain games straight up dislike running on them if I don't tune core affinity and memory access settings.  I do hope that IO die does help these new chips.  If it does, AMD better get a new TR out with it (plus, put it on the 7nm+ EUV as well please).

Barring any hidden latency improvements, the latency should be a mixed bag. There will be no NUMA. Near latency should be worse but far latency should be much better. To clarify: those terms should not exist on Zen 2 but we're comparing to the previous products which did have near and far latencies. The latencies should be uniform and consistent hence no NUMA or NUMA-like shenanigans but there should be a penalty for moving memory controllers off die but it'll be easier on software and overall the trade-off should be good. The bigger caches should also help nullify the performance deficit from any added latency.

 

Of course this is speculation because we haven't seen any hard data on this and there might be some curve balls in there.

Link to comment
Share on other sites

Link to post
Share on other sites

If the 16 core will be same as 9900k when it comes to single core, I'll be saying goodbye to my golden 9900k sample...

Main PC:

CPU: Intel Core i9 13900KS SP 116 (124P-102E) (6.1Ghz P-Cores 4.8Ghz E-cores) MC SP 88

CPU Voltage: LLC8 1.525V (real voltage 1.425V + - Temps 85-90 P-Cores, 70-73 E-cores)

Cooled by: Supercool Direct Die 14th gen full nickel

Motherboard: Z790 ASUS Maximus Apex Encore

RAM: GSkill TridentZ 2x24GB DDR5 8600Mhz CL38 (OC from 8000Mhz CL40)

GPU: RTX MSI 4090 Suprim X with EKWB waterblock

Case: My own case fabricated out of aluminium and wood

Storage: 4x 2TB Sarbent Rocket Plus Gen 4.0 NVMe, 1x External 2TB Seagate Barracuda (Backup)

WiFi: BE202 WiFi 7 Tri-Band card module

PSU: Corsair AX1600i with custom black and red cables with 2x Corsair 5V+ Load Balancer

Display: Samsung Oddysey G9 240Hz Ver. 5120x1440 with G-Sync and Freesync Premium Pro 1008 Firmware Ver, and 1x Electriq USB C 1080p 15'8 inch IPS portable display for temperature and stats, MSI 23'8 144Hz G-Sync

Fan Controllers:  6x AquaComputer Octo with 5 temperature sensors

Cooling: Three Custom Loops:

1st Loop: 5x 480mm XE CoolStream radiators with 1x Revo D5 RGB pump and 1x Rajintek Antila D5 Evo RGB pump for GPU only cooling with 2x Koolance QDC3, red coolant

2nd Loop: 5x 480mm XE CoolStream radiators with 1x Revo D5 RGB pump and 1x Rajintek Antila D5 Evo RGB pump for CPU only cooling with 2x Koolance QDC3, purple coolant

3rd Loop: 1x 240mm PE CoolStream radiator with 1x EKWB Revo D5 pump (RAM ONLY)

Total: 5x pumps and 13x radiators 50x 3000RPM Noctua Industrial fans

Keyboard: Razer BlackWidow V3 RGB - Green switches

Sound: Logitech Z680 5.1 THX Certified 505W Speakers

Mouse: Razer Basilisk Ultimate Wireless with charging dock

Piano: Yamaha P155

Phone: Oppo Find X5 Pro

Camera: Logitech Brio Pro 4K

VR: Oculus Rift S

External SSD: 256GB Overclocking OS

LaptopMSI Titan GT77HX V13RTX 4090 175W, i9 13980HX OC: P-Cores 5.8Ghz 3 cores and 5.2Ghz 5 cores and E-Cores 4.3Ghz, 192GB of RAM @5600Mhz @3600 (chipset limit),

12TB (3x4TB) of NVMe, 17'3 inch 4K 144Hz MiniLED screen, 4x 17'3 ASUS portable USB-C Monitors 240Hz, Creative Sound Blaster G6 Sound Card, Portable 16TB NVMe in TB4 enclosures (8x2TB), Razer Basilisk Ultimate Wireless with charging dock gaming mouse, Keychron K3 gaming keyboard with blue switches low profile, Logitech Brio 4K Webcam.

Hand held: ROG Ally with XG Mobile RTX 3080 with Keychron K3 low profile keyboard (Blue Switches) and Razer Hyperspeed V3 mouse and 4TB NVMe upgrade (WDBlack SN850X), with 100W 20000Mah power bank and portable monitor ROG XG17AHP 17'3 inch 240Hz with built in battery, and 518Wh Power station for Camping.

Link to comment
Share on other sites

Link to post
Share on other sites

I think that 4.8ghz would probably be possible. And just like previous zen cpus all of them should be overclockable to get to that point, its just amd artificially making lower clock speeds from factory to sell at different price points. Thats how the last 2 zen were correct? If so then im 99% sure this is how itll be. 

Link to comment
Share on other sites

Link to post
Share on other sites

3700x to 3800x appears to be binning and power target.

3900x only has the 1 sku (at this time at least).

 

There's a few ways to look at how this can unfold, depending on if you're skeptical or optimistic.

 

The skeptical view is that the binned 8 core chiplets on the 3800x are as good as they could get, and to go higher clock speeds like they wanted for the 3900x, they had to kill off 2 of the cores in the chiplet, with those that couldn't cut it being declocked and turned into the 3600 or 3600x.  Thus, they're not shipping a 16 core chip because they couldn't keep the clock speed up.

 

The optimistic view is that the 3900x are the 8 core chiplets from the 16 core confirmed to exist that couldn't quite cut an even  higher clock speed, so got demoted and cores turned off.  The 3900x doesn't have a non-x part listed, so this isn't quite as far of a stretch to think as I'd normally say, but would require some serious binning, and end up with the 8 core chiplets for the 16 core being in very short supply, and demanding a price to match.

Link to comment
Share on other sites

Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now


×