-
Competitor rules
Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.
Ryzen "2" ?
- Thread starter beany_bot
- Start date
The Intel AVX512,Venn diagram of confusion!
https://twitter.com/instlatx64/status/870285161979682818
Yup, to decipher it:
CannonLake
AVX512VBMI
AVX512IFMA
AVX512BW
AVX512DQ
AVX512VL
AVX512F
AVX512CD
SkyLake Xeon Core-X
AVX512BW
AVX512DQ
AVX512VL
AVX512F
AVX512CD
Knights Landing
AVX512F
AVX512CD
AVX512ER
AVX512PF
Knights Mill
AVX512F
AVX512CD
AVX512ER
AVX512PF
AVX512_4FMAPS
AVX512_4VNNIW
AVX512VPOPCNTDQ
Soldato
I want the new CPU's to do well but I can't see them getting any faster than 4.2 or 4.3.
8 core at 4.3 isn't to be sniffed at though if the pricing is good then INTEL will have to drop their prices too.
8 core at 4.3 isn't to be sniffed at though if the pricing is good then INTEL will have to drop their prices too.
Caporegime
My view is they don't need to, what are we comparing them to here? the 8700K? that's fine, 6 cores 12 threads at 5Ghz, great, £350.
Ryzen 5 2600 With the same IPC as the 8700K but 4.3Ghz, £190, is that not also great?
Soldato
Ah but IS the IPC going to be the same? If it is then brilliant, AMD have nailed it this time around.
Caporegime
What do you think the IPC difference is now? Cinebench R15, Ryzen 5 1600 at 4Ghz scores about 1350, 8700K at 5Ghz about 1700, that's a 26% difference in score vs 25% difference in clock speed.
Ryzen 1 vs Coffeelake IPC is as near as makes no difference identical.
Last edited:
IPC is pretty close already in most cases. Clock speed is the biggest drawback in gaming and a few other applications.
Caporegime
Yes, i'm not being funny but i just went from a 4690K @ 4.5Ghz to a Ryzen 1600 @ 3.9Ghz and i have yet to find a game where the Ryzen CPU is not faster, i still play a lot of old games with mates because they are a huge laugh, Arma III, Killing Floor II, Insurgency, Borderlands 2 / Pre-Sequel, we have just got back into blasting around in Serious Sam, in everything my frame rates are higher, this on a GTX 1070 already pulling deep into the 100's of FPS and despite a 600Mhz deficit.
The only old game like that i do have with a built in benchmark was Dirt Showdown, its 20% higher Frame rates with the 3.9Ghz 1600 vs the 4.5Ghz 4690K.
Vs Coffeelake Ryzen just needs more Mhz...
The only old game like that i do have with a built in benchmark was Dirt Showdown, its 20% higher Frame rates with the 3.9Ghz 1600 vs the 4.5Ghz 4690K.
Vs Coffeelake Ryzen just needs more Mhz...
Last edited:
Soldato
Hands up, I didn't know the difference between the two.
Caporegime
Few people do, to be fair most people, including mainstream reviewers still say "Intel has higher IPC" JayZ2Cents whom i respect and like as a reviewer still says it.
To be fair to them in some cases that is true, if you weigh up the single threaded performance in the same Cinenbench you will find Coffeelake is about 8% better, AMD's better than Intel's SMT lets them close that gap to 0 in MT, but again this is still an Intel friendly benchmark, it always was and still is, there are other high load applications where Ryzen is 'significantly' faster than Coffeelake and others where Coffeelake is significantly faster.
If those with the million subscribers actually took the time to look at the IPC question objectively they would find that using a broad brush generalization is wrong, in some instances yes Ryzen has lower IPC, in others Ryzen has higher IPC, mostly they fall somewhere in between, they are different CPU's, one simply cannot broad brush them.
With Mainstream reviewers doing it, what chance does Joe have of knowing whats factually correct?
On the question of gaming. When you actually look at game benchmarks, lots of them, not just one or two outliers what you find is the performance difference is almost always equal to the clock rate differences, sometimes that performance difference is even less than the clock rate difference.
Soldato
Ryzen does not have coffeelake IPC, it is closer to haswell than any other intel architecture. It comes close in certain tests like cinebench but when a larger sample of testing is done you can see the differences.
Below are the results of a large testing package in a core vs core frequency vs frequency test.
People would do well to read this article https://forums.anandtech.com/threads/ryzen-strictly-technical.2500572/
Below are the results of a large testing package in a core vs core frequency vs frequency test.
People would do well to read this article https://forums.anandtech.com/threads/ryzen-strictly-technical.2500572/
Caporegime
Ryzen does not have coffeelake IPC, it is closer to haswell than any other intel architecture. It comes close in certain tests like cinebench but when a larger sample of testing is done you can see the differences.
Below are the results of a large testing package in a core vs core frequency vs frequency test.
People would do well to read this article https://forums.anandtech.com/threads/ryzen-strictly-technical.2500572/
Absolute IPC based on what? given that Coffeelake and Ryzen have different IPC levels compared to each other based on workload / application the very concept of "Absolute IPC" is utterly flawed.
He doesn't even say what application/s? he used to draw his flawed blanket statement from.
If i use Corona benchmark, World Machine, Algorithmic Substance applications, Linux based Compilers or cherry picked workloads in Blender.... i could very easily trump up a chart like that and claim "Absolute IPC" with Ryzen topping the chart.
The fact of the matter is my doing that and thus his chart is Male Bovine Manure. he obviously didn't use Cinenbech or include it to come up with that chart, if he had it wouldn't look anything like that, despite using it to measure power consumption.
He's cherry picking to fit a predetermined hyperbole click-bait outcome,
any smuck can do that.
Last edited:
Soldato
Read the article, it shows you.
This is the reason why claims like "ryzen has coffeelake ipc" are just dumb. You used cinebench to claim this, thestilt used a crap ton more tests to produce that chart.....
It is ALL dependent on the application in question.
To say the stilt has cherry picked the charts is also a little silly, he is one of the reasons people are actually able to hit ram speeds with ryzen. Asus even implemented his settings into their bios' for God's sake.
Caporegime
I did read it, the first few times you posted it, it does not say how he got to those conclusions.
This is what i said and used it to make it perfectly clear that using language like "Absolute IPC" makes it completely flawed, the fact that you also used it back at me suggest you understand that picking out individual applications to determine an "Absolute IPC" is a contradiction, its flawed. I repeat.
He's a troll.
Soldato
I suggest you do some research on who he is and what he has done for AMD before calling him a troll.
Soldato
If AMD has lower performance in games based on older engines,is not really an issue with IPC or even clockspeeds. In non-gaming applications Ryzen is generally higher up against Intel CPUs as opposed to purely gaming orientated ones. Its mostly down to games devs not supporting their games properly and updating them to run better on newer CPUs. It was the same issue with Skylake X. I mean look at an AMD partner like Bethesda. They CBA it appears. They are more worried about VR and "mini-DLCs" instead of patching their open world games they actually make(not all the ones they publish and not make themselves).
I think its probably more like the Venn diagram of despair if you are a software dev!
Yup, to decipher it:
CannonLake
AVX512VBMI
AVX512IFMA
AVX512BW
AVX512DQ
AVX512VL
AVX512F
AVX512CD
SkyLake Xeon Core-X
AVX512BW
AVX512DQ
AVX512VL
AVX512F
AVX512CD
Knights Landing
AVX512F
AVX512CD
AVX512ER
AVX512PF
Knights Mill
AVX512F
AVX512CD
AVX512ER
AVX512PF
AVX512_4FMAPS
AVX512_4VNNIW
AVX512VPOPCNTDQ
I think its probably more like the Venn diagram of despair if you are a software dev!
Also remember this is release-day testing, prior to any Ryzen patches and prior to Meltdown/Spectre. Probably not as accurate now.
I also don't think he mentions what RAM speed he was using? At launch it's not likely to be 3000+ MHz.
Soldato
If AMD has lower performance in games based on older engines,is not really an issue with IPC or even clockspeeds. In non-gaming applications Ryzen is generally higher up against Intel CPUs as opposed to purely gaming orientated ones. Its mostly down to games devs not supporting their games properly and updating them to run better on newer CPUs. It was the same issue with Skylake X. I mean look at an AMD partner like Bethesda. They CBA it appears. They are more worried about VR and "mini-DLCs" instead of patching their open world games they actually make(not all the ones they publish and not make themselves).
I think its probably more like the Venn diagram of despair if you are a software dev!
Just take a look at the article I posted, ryzen REALLY struggles in some of the tests he ran. It might go some way to explaining some of the strange gaming results.
Linpack/open SSL/himeno/nbody are the ones that stand out by a mile. Bear in mind these are at the same core count and frequency.
Permabanned
What have you found so far Gavin? I mean you have both a 1700 and a 8700k, why don't you limit your 8700k to the same clockspeed and memory speed/timings and report back to us with some real world testing.
Caporegime
I'll explain it then.
Ryzen performance differs depending on what its doing, actual benchmarcks on the Internet show its IPC to be between and at Haswell and Coffeelake.
So let me put it to you this way.
when you have a bucket of performance testing and their results, you dip your hand into said bucket and use those results as a medium for your own combined results.
If your combined result is equal to worst case scenario, like 'no better than Haswell' then maybe you have just been very unlucky with your blind result harvesting.
Be that as it may you would at least acknowledge that these results cannot be absolute indicative of the products performance.
Now, do you know what a straw man is? have you ever hear that term? its where someone takes something that has a smidgen of truth, is true sometimes or can be, a straw-man takes those outliers and frames them as absolutes.
What the author in your link did was write something that can leave the reader in no doubt that Ryzen is no better than Haswell, whats more he then not only frames it as an absolute he actually said and used the words "Absolute"
He is a blindingly obvious troll if not an outright shill.