July 11, 20196 yr Author Commercial Member To be honest 28 minutes ago, westman said: Intel 9900k all core boost is 4,7ghz , 5.0ghz thats on single tread app , not in the sim have you read the Intel factory specs or only locked at max single core boost like amd 4.6 or was it 4.7ghz thats correct 5.0ghz Iam very impressed of the IPC AMD have manged, clock for clock wise cross my fingers that they can mange 4.8 all cores soon To be completely honest thats what google told me. 🙂 I am nothing but a glass half empty AMD CPU fan-boy It’s so nice to see AMD'd cup running over, that my smugness is leaking out. The IPC on AMD is amassing, Linus did a great review on all the features of the new Ryzen CPU's and the whole time I am thinking, how much of a leap this is from bulldozer, they literally leap frogged right over Intel. And there is so much more potential in this chip. And for some reason Intel just can’t catch up. AMD use to be the CPU you had to try and justify, now those tables have turned. Edited July 11, 20196 yr by Nagmaal
July 11, 20196 yr 50 minutes ago, Nagmaal said: To be honest To be completely honest thats what google told me. 🙂 I am nothing but a glass half empty AMD CPU fan-boy It’s so nice to see AMD'd cup running over, that my smugness is leaking out. The IPC on AMD is amassing, Linus did a great review on all the features of the new Ryzen CPU's and the whole time I am thinking, how much of a leap this is from bulldozer, they literally leap frogged right over Intel. And there is so much more potential in this chip. And for some reason Intel just can’t catch up. AMD use to be the CPU you had to try and justify, now those tables have turned. rember back in the AMD golden days it was the same Intel have the FRQ but poor ipc I run FX54 and FX57 that outperformed Intel with much lower FRQ. It would be nice to see that one more time , they ar close but not there yet if you only count raw performance, price performance they win easy. http://
July 11, 20196 yr Commercial Member Used to be GHz alone wasn't a good enough metric, now it is maybe getting worse, not just from the usual floating point hardware and matrix handling performance gains. There's preemptive memory addressing reordering and alignment, concurrent interleaving between cores to a degree and other techniques. To what extent these things represent increased performance, especially in multi core software is not clear. Things can be a problem on one chip and less to another that we don't expect. Since these chips don't clock so much as they sing, they create harmonics, scaling frequencies causes implications. Data is scrambled on the bus to decrease interference depending on the types of data so that's work to be done too. Steve Waite: Engineer at codelegend.com
July 11, 20196 yr Check out the post today on Tomshardware don't expect the 3000 series to run at max boost on all cores. Raymond Fry.
July 12, 20196 yr Author Commercial Member Sorry but you have lost me litle. Thats been a Ryzen thing since the 1st Generation. You need single core performance, you use the strongest clocking Core, you need multi threaded performance, well Ryzen is a beast at multi treading. Edited July 12, 20196 yr by Nagmaal
July 12, 20196 yr Author Commercial Member 15 hours ago, westman said: rember back in the AMD golden days it was the same Intel have the FRQ but poor ipc I run FX54 and FX57 that outperformed Intel with much lower FRQ. It would be nice to see that one more time , they ar close but not there yet if you only count raw performance, price performance they win easy. I fondly remember those days, all my friends had Pentium 4s and here comes me with my Athlon 64 4000+ , a cheaper processor but outperforming them. It was so hard to not be smug about it. 🙂
July 12, 20196 yr The price thing is no longer a factor you can pick up a 9900k for less than a 3900X in the UK, A 5ghz capable CPU instead of a 4,3 boost if you are lucky according to feedback no AMD CPU is hitting the stated boost AMD blame this on the silicon lottery, on the post on TomsHardware manual overclocking is futile. Raymond Fry.
July 12, 20196 yr Commercial Member On 7/11/2019 at 10:36 AM, SteveW said: Used to be GHz alone wasn't a good enough metric, now it is maybe getting worse This. Comparing CPUs of different architecture on clock speed is like comparing auto engines by redline. Interesting, but rarely useful. Cheers! Luke Kolin I make simFDR, the most advanced flight data recorder for FSX, Prepar3D and X-Plane.
July 13, 20196 yr 20 hours ago, Luke said: This. Comparing CPUs of different architecture on clock speed is like comparing auto engines by redline. Interesting, but rarely useful. Cheers! Excellent point! My computer: ABS Gladiator Gaming PC featuring an Intel 10700F CPU, EVGA CLC-240 AIO cooler (dead fans replaced with Noctua fans), Asus Tuf Gaming B460M Plus motherboard, 16GB DDR4-3000 RAM, 1 TB NVMe SSD, EVGA RTX3070 FTW3 video card, dead EVGA 750 watt power supply replaced with Antec 900 watt PSU.
July 15, 20196 yr Author Commercial Member The new Ryzen CPU's have a better IPC (Instructions per clock) So Clockspeed is a not a very good measure of performance. It was mentioned on Linus tech tips that a Assus released a patch on Saterday before release that solved some overlclocking issues. I dont know how many reviewers reran their results before they Published on Sunday , but they did.
July 15, 20196 yr Der Bauer has re-tested the 3900X on the latest 1003 sent to him by AMD and even on a AIO cooler you will not see 4.6ghz in games or anything check it out, Gamers Nexus did an OC over 4.6 with liquid Nitro. Raymond Fry.
July 15, 20196 yr Author Commercial Member 1 hour ago, rjfry said: Der Bauer has re-tested the 3900X on the latest 1003 sent to him by AMD and even on a AIO cooler you will not see 4.6ghz in games or anything check it out, Gamers Nexus did an OC over 4.6 with liquid Nitro. We are not comparing performance based on GHZ. It does not matter when one CPU has better IPC (Instructions per Cycle), its about efficiency. When the new Ryzen CPU does more instructions per cycle, GHZ is not a measure. Imagine 2 cars one making 500hp the other making 440 Horsepower. Witch one is faster. The 500hp one? What if the 440hp one weighed half of what the other weighed. Horsepower is no longer a valid measuring device. Benchmarks become the measuring stick. You have to run Benchmarks. Benchmarks says the new Ryzen is the obvious choice and Intel is the Choice you have to justify. Edited July 15, 20196 yr by Nagmaal
July 15, 20196 yr 2 hours ago, Nagmaal said: We are not comparing performance based on GHZ. It does not matter when one CPU has better IPC (Instructions per Cycle), its about efficiency. When the new Ryzen CPU does more instructions per cycle, GHZ is not a measure. Imagine 2 cars one making 500hp the other making 440 Horsepower. Witch one is faster. The 500hp one? What if the 440hp one weighed half of what the other weighed. Horsepower is no longer a valid measuring device. Benchmarks become the measuring stick. You have to run Benchmarks. Benchmarks says the new Ryzen is the obvious choice and Intel is the Choice you have to justify. Simple then don't OC your CPU for flight sim it uses less power your choice, if however you want to do video editing you may be better with threadripper as the prices are tumbling. Raymond Fry.
July 15, 20196 yr Commercial Member When these simulators are tested and benchmarked, it's not straightforward and I don't pay any attention to them whatsoever. Steve Waite: Engineer at codelegend.com
July 15, 20196 yr Commercial Member 6 hours ago, Nagmaal said: The new Ryzen CPU's have a better IPC (Instructions per clock) As I already mentioned comparing IPC is a waste of time. We have concurrent out of order memory transactions just one thing to throw that off, you need the data ready to use the instructions on. Edited July 15, 20196 yr by SteveW Steve Waite: Engineer at codelegend.com
Archived
This topic is now archived and is closed to further replies.