Here is my analysis on the Ryzen 3000 1) Looks like AMD didn't get as high clocks as they would've liked. Seems like the 7nm HPC right now can get 4.4 GHz very consistently but beyond that requires much more power and chiplet binning. Proof? Look at 3800X. It requires 40 watts and $70 more for a mere 100 Mhz boost. I'm guessing the $70 premium is for binning. Wouldn't be surprised if 3700X can only OC to 4.45 Ghz or thereabout all core. 2) No 5 GHz. Intel would've been fucked if AMD could reach 5 GHz which I'm guessing they will as 7nm process matures. 3) IPC uplift. This is the only thing keeping AMD in game. Otherwise, 7nm didn't give them much advantage except 100-200 Mhz boost. 4) No 16 core variant. WTF? When did AMD become this jewey. Everyone was waiting for this one. 5) They showed many benchmarks but they all favored AMD so I'm still a little cautious. I'm guessing an OC'd 5Ghz 9900K will still be faster than OC'd 3800X. That's why they have positioned 3900X at 9900k pricepoint so that it becomes a no brainer to buy 3900X.
>Inturd and AMDtards fanboys and shills keep out plz >"inturd security lelelel" idiots fuck off
>Proof? Look at 3800X. It requires 40 watts and $70 more for a mere 100 Mhz boost. I'm guessing the $70 premium is for binning. Imagine being such a brainlet
Wyatt Lopez
GTFO and Cope, Shill.
Hunter Gonzalez
>no answer >brainlet/shill Expecting AMDtards to be able to read was too much I guess.
Brody Foster
I think it’s a fair analysis. I thought the 3850X was supposed to be their 16 core variant, is it not going to be announced?
Zachary Cox
>No 16 core variant That's threadripper territory, they would cannibalize their workstation market if they were to release a 16 core ryzen and I'm not even sure they can fit 16 cores into the ryzen package, that 12 core already seems cramped with dies.
Not sure why they keep quiet about threadripper tho.
Carter Reed
>Otherwise, 7nm didn't give them much advantage except 100-200 Mhz boost. It gave them +250MHz boost and 1.5x the core count just by comparing the 3900x vs 2700x
>They showed many benchmarks but they all favored AMD so I'm still a little cautious From what I understand, zen+ lost at pubg so I wouldn't say this is the case. It could be cherry picked by AMD due to it being an outlier though.
>I'm guessing an OC'd 5Ghz 9900K will still be faster than OC'd 3800X Possible. We'll know by early July
Aiden Collins
zen 2 4.4ghz all core overclock should be about as fast as a 5ghz all core 9900k
William Bennett
To point 1, it matters a lot just how steep the voltage curve is. There is a world of difference between not being able to overclock much past stock clocks and it just taking tons of power and cooling buy still being possible. We aren't going to know the slope of the voltage curve in this region until people start trying to overclock review samples.
Also don't forget that base and all core boost affect TDP for AMD, so looking at the maximum stock boost difference doesn't tell anywhere near the whole story on the TDP difference.
2+3) 15% IPC is quite a lot in one year. 7nm may give somewhat more overclocking headroom as well,depending on the slope of the voltage curve compared to GF 14nm and 12nm. 7nm also gives the possibility of going to 16 cores on a mainstream socket at little to no clock tradeoff.
5) I don't see how Cinebench R20 favours AMD at all, nor the array of games they compared performance on. Again, until we know the voltage scaling there is little point in trying to guess OC vs OC.
Adam Martinez
does the X570 chipset just add PCIe 4? or will the B450/X470 also have support for it with some update?
Jordan Thompson
x570 has chipset level pcie4. x470 and older could support it, but only on the direct cpu lanes.
Tyler Howard
do i geet 3600 or 3600x 3600x seems way better since its 95w big oc headroom i guess
as always it didnt live up to the hype. 9900k will slaughter it in 1440k
Samuel Anderson
BASED.
Jeremiah Wilson
stock tdp has nothing to do with overclocking headroom the 3600x is probably binned better so you might get 200mhz more out of it
Ian Cox
>MUH FREQUENCY >completely ignores the 15% IPC uplift shill thread
Cooper Hill
inturd security lelelel
Charles Price
> 15% IPC improvement > Obsessed with 5.0 GHz meme number
Do you even understand what you're talking about? That IPC improvement is nuts, the 3700X and 3800X are tied with the 9700K and 9900K at *lower* frequencies because of it, all while drawing less power.
This throws the entire 9th generation in the dumpster. Same or better performance at a lower price.
Jaxon Evans
I'd say the 3600X, although it's not about overclocking since XFR and PB will do that automatically, better than a manual overclock 90% of the time.
Anthony Davis
thats definitely not the resolution where it will slaughter zen 2
Gavin Barnes
>No 5 GHz. Intel would've been fucked if AMD could reach 5 GHz which I'm guessing they will as 7nm process matures. We don't know how good it overclocks. Amd respect tdp while stock (((95w))) 9900k consumes 200w under load.
Jonathan Barnes
My Gigabyte MoBo has already received a Bios update for PCIe 4.0 but it's only for the first 16x lane.
Isaiah Rivera
>AMDrones now care about power consumption despite running AMD GPU housefires since a decade
Adrian Stewart
was there any word for when it will be released and/or when 3rd parties can get it to review and bench
Nathan Wright
That's not what i wrote, you humongous faggot, it's about following advertised spec, which also was the case with fx with 200w cpu consuming 200w.
Brody Gray
July
Angel Butler
Here's the answer: Everyone but you knows that listed TDP numbers are complete bullshit and mean nothing.
Brody Clark
>Not sure why they keep quiet about threadripper tho. Their pondering on changing the name to Threadraper.
Ryder Bell
>Not sure why they keep quiet about threadripper tho. They might just not want to completely Osborne the current Threadrippers.
Michael Reyes
That is fucking hilarious, corporate banter is getting good these days.
Joseph Lopez
release or tech "journos" getting early access?
Hudson Cox
TDP ≠ Watt
Christopher Bell
>It requires 40 watts and $70 more for a mere 100 Mhz boost. I don't think "requiring 40 W mre" is true, the TDP is far more likely to be about binning. In order to reach the higher clock speeds, they probably have to use "some" chiplets that don't do as well in efficiency, and therefore they've reserved a much higher TDP than the majority of them will need.
Does win7 work on current gen ryzen? can I expect X570 motherboards to have Win7 drivers?
sandy bridge fag here wanting to upgrade but want to stick with win7 fuck win10
Caleb Thompson
It takes slipstreaming the USB drivers onto the Windows 7 install media, but yes you can install windows 7 to a Ryzen system. Whether that can still be done when Zen2 drops remains to be seen.
Andrew Diaz
>Does win7 work on current gen ryzen? Not by default, but you can do some fuckery with drivers to make it work. But IIRC windows required some patches to unfuck its scheduler and make it utilize ryzens properly, so it still won't be working at full power.
Brody Allen
I wanna see some 3rd party benchmarks. Looks promising, but I still wont get my hopes up.
Jason Bennett
>>"inturd security lelelel" idiots fuck off why? >SECURITY DOESN'T MATTER enjoy your NSA/Mossad backdoor
Eli Campbell
>serious thread >HURR DURR AMD CANT GIVE 5GHZ AND LEAVE NO ROOM FOR OC THEREFORE ITS SHIT >AMD BINS CHIPS >ONLY THING THAT SAVES AMD IS IPC HURR DURR >NO 16 CORES HURR DURR EVERYONE WANTED A NICHE PART BECAUSE 1950X WAS SUCH A HUGE SELLER AFTERALL >HURR DURR AVX R20 NOW FAVORS AMD >VECTORING DOESNT MATTER ANYMORE
Charles King
You're looking at TDP as a means to measure power consumption. And that's not totally the case. Especially with Intel. Wait and see how much power is actually consumed at what boost clocks. Anything else is pointless Naval gazing.
William Scott
1. use mainboard manufacturer Win7 ISO Tool to include USB3 and NVMe driver 2. install win7 unfucker to still get updates It's that easy to install win7 on Ryzen/Intel 9XXX
Matthew Watson
The 12 core is a 16 core with 2 cores per chiplet disabled.
Liam Garcia
this. OP has no idea what he's talking about. Embarassing. Clearly a vkiddo
Angel Cruz
>The 12 core is a 16 core with 2 cores per chiplet disabled.
lower binned chiplets to enhance yields, most likely to be unlockable
intel is done for
Jace Jones
My guess is that they don't boost to 5ghz only because the AM4 socket and the current mobos aren't rated for 220Watts.
But i think manual OC could get to 5ghz or really close.
Nolan Ramirez
AMD is CLEARLY sandbaging. >1) Looks like AMD didn't get as high clocks as they would've liked. Seems like the 7nm HPC right now can get 4.4 GHz very consistently Why would undermine there profit margins when they when they already beat Intel in single threaded performance? The best dies go to Epyc which is in EXTREMELY high demand. Except significant clock bumps from the Zen 2+ refresh as 7nm yields improve. >2) No 5 GHz. Intel would've been fucked if AMD could reach 5 GHz which I'm guessing they will as 7nm process matures. Didn't need to. >3) IPC uplift. This is the only thing keeping AMD in game. Otherwise, 7nm didn't give them much advantage except 100-200 Mhz boost. Are you looking at the same TDPs i am? >4) No 16 core variant. WTF? When did AMD become this jewey. Everyone was waiting for this one. When there enterprise customers where willing to buy those dies for $4k a pop. AMD isn't a charity. >5) They showed many benchmarks but they all favored AMD so I'm still a little cautious. The new cinebench heavily favors Intel.
Either sell 16c consisting of 2 perfect ZEN2 dies to gaymerkids that have no use for 16 cores for ~$650~ --OR-- make BIG money from EPYC which requires all their flawless dies. we might see 16c in a few months when they've stockpiled enough lower binned 8c dies. 3700x is the CPU to go for non-brainlets
Thomas Howard
post a smaller image
Aiden Bennett
that is a funny way to spell 120p resolution
Blake Walker
>Looks like AMD didn't get as high clocks as they would've liked.
says who? rumor mill garbage? ryzen 3000 clocks way better than 1000 and 2000 chips. not counting architectural improvements that increased IPC itself.
>no 5ghz
again only rumor mill garbage, zen 2 at 4.5 ghz literally has the same performance as coffee lake with 15% less clock speed and half the power. Sure its not as good for flexing your epeen size but efficiency and value is what OEMs want, and OEMs are where intel and AMD both make most of their money
>IPC uplift. This is the only thing keeping AMD in game
zen + literally already has better IPC than coffee lake clock for clock, and AMD pulled another 15% out of their hat. Your summerfagness is painfully obvious as this is nothing short of amazing considering that for the last 6 years, since ivy bridge, it was largely thought that double digit architectural IPC gains were a thing of the past.
>No 16 core variant. WTF?
Its called strategy, intel cant currently compete with the 2800x let alone the 2900x for mainstream computer CPUs. Its better for AMD to sells less skus right out of the gate as demand is going to be really high july 7th, and epyc is doing really well as it is, so having to bin even more chiplets would be a waste of money just to have a halo product without any competition anyways. And besides when the 16 core chip comes out in response to the cannon lake 10 core, it just give amd another chance for dumbshits to upgrade to it from their 3900x.
>I'm guessing an OC'd 5Ghz 9900K will still be faster than OC'd 3800X
9900k already is clocked at 5ghz from factory, and with a custom loop and some luck you might see 5.2 ghz out of it. Assuming the 3800x cant overclock at all, the 9900k would be a whole 5% faster but will cost a whole $400 more just in the price of the chip and the cooling system you'll need for those extra 200mhz.
Hudson King
7/7 general release, and maybe reviews before that from lapdog outlets that are on a short leash
Colton Garcia
could some of you fags do something similar to this but with intel/2nd gen ryzens i find it interesting how good ryzen is at lower clocks/voltages apart from my oc i've seen 1(one) thing like this and it was in russian
I'm in a similar boat, though ivy bridge here. Gotta get my hands on a legit win7 ult serial for a change, since I can't be bothered doing a bios mod this time around given it was a headfuck to redo after I'd updated the bios years later and forgotten everything.
Doesn't that still partly rely on the motherboard manufacturer actually providing drivers for shit, or would you just go look up every individual component that can have drivers and download them from elsewhere/just use the win10 variant on 7?
Angel Sanders
holy kek
Julian Murphy
No it's fucking cringey made to attract easily humored morons into buying their processors
Joseph Perry
>The new cinebench heavily favors Intel. Citation needed.
Levi Allen
Of all the words you could have used you have to use the laziest word possible, the most blanket and least meaningful of all words you could have used.
