AMD's flagship Threadripper CPU has appeared on Geekbench

So the single-core score is closest to the four-year old Intel Core i7-4770R and the multicore is a little lower than an Intel Core i7-6900K (which cost about £835) today.

Not impressed yet not surprised.
 
I live in hope that one day AMD will sort out their disappointing single core IPC.

There is nothing wrong with their IPC. It is fantastic. They need to make it clock better is all.

So the single-core score is closest to the four-year old Intel Core i7-4770R and the multicore is a little lower than an Intel Core i7-6900K (which cost about £835) today.

Not impressed yet not surprised.

That's not right. We all know that Ryzen is around Broadwell E or better. Something is seriously not right with that benchmark.
 
It's still two generations behind Intel's best.

So what? are they charging Intel money for it?

Last time I looked the 1700 was around £280 for the same performance as a £1000 Intel CPU.

It's not about who can wee the highest.
 
The cost is a completely different metric.

There are people like me who are willing to pay more money for the superior IPC performance of Intel.

...But if AMD could offer a CPU with similar IPC, would have no problem in switching to AMD from Intel.

AMD are missing out on sales with inferior IPC architecture.
 
Last time I looked the 1700 was around £280 for the same performance as a £1000 Intel CPU.

The cost is a completely different metric.

There are people like me who are willing to pay more money for the superior IPC performance of Intel.

...But if AMD could offer a CPU with similar IPC, would have no problem in switching to AMD from Intel.

AMD are missing out on sales with inferior IPC architecture.

So you'd be willing to pay more than £700 for just "some" superior performance? :mellow::huh:
 
The cost is a completely different metric.

No it really isn't. Not unless you have far too much money and no sense. The cost is everything on everything. Everything has a price, it just depends if you have enough money to ignore that and not care about it.

For example I have not upgraded from my Titan XM because I don't think the prices are right. Nothing to do with whether I can afford it or not but much more to do with me simply refusing to pay the price Nvidia are asking. If they were not being so ridiculous then I would have gone 1080/Titan XP or 1080Ti for sure.

As an example on launch you could pick up a basic 1070 for £360 or so. Now? they are over £400. Yet a year has passed, so how does that work out?

^ just a basic example for you. If my Titan XM packed in I would have to pay more than I paid for it for the same performance. Something doesn't add up there.

There are people like me who are willing to pay more money for the superior IPC performance of Intel.

Then good for you. But what part of that means you have to go around belittling every one else? Oh wait, that's a part of it for you, right?
 
Something iffy with that multi core score, looks similar to some 1800x results. With that single core result and how Ryzen scales the multi core score should be a lot higher.

Going to put that anomaly to either software or ram speed (at 1065mhz any Zen chip is going to be severely nerfed in multi core tests, given how they interact) assuming the leak is accurate.

In relation to your complaints about single core performance Bagpuss, really not sure what that is all about. From what i've seen current AMD cpu's are barely 5% behind Intel offerings in terms of IPC. Not bad considering that they've made up for it with cost and core count. Kaby Lake's only significant advantage is clock speed, and to be honest that's only the case because of how much they've refined the architecture. Ryzen was never going to hit 5ghz in its first iteration, just like kaby lake didn't when it was skylake.
 
Something iffy with that multi core score, looks similar to some 1800x results. With that single core result and how Ryzen scales the multi core score should be a lot higher.

Going to put that anomaly to either software or ram speed (at 1065mhz any Zen chip is going to be severely nerfed in multi core tests, given how they interact) assuming the leak is accurate.

In relation to your complaints about single core performance Bagpuss, really not sure what that is all about. From what i've seen current AMD cpu's are barely 5% behind Intel offerings in terms of IPC. Not bad considering that they've made up for it with cost and core count. Kaby Lake's only significant advantage is clock speed, and to be honest that's only the case because of how much they've refined the architecture. Ryzen was never going to hit 5ghz in its first iteration, just like kaby lake didn't when it was skylake.

Not to mention very few people actually overclocked their 7700K's to 5Ghz. Just like very few people are going to be and overclock a 7900X to 4.8Ghz, contrary to what der8auer suggests. This obsession with clock speed is more about what the CPU can theoretically do, not what it realistically is going to do. It's more of a tagline—ironically one that Intel prefers not to mention since Intel asked 7700K users to not overclock and since der8auer discusses delidding, something that should not be necessary in a high-end product and is not supported by Intel—to sell chips to people who only use €40 air coolers and want stability.
 
So the single-core score is closest to the four-year old Intel Core i7-4770R and the multicore is a little lower than an Intel Core i7-6900K (which cost about £835) today.

Not impressed yet not surprised.

Considering an 1800X performs around the same as a 7700K, Surpasses it sometimes in various games, And performs around the same as a 6900K in multitasking, I'd take these benchmarks with a grain of salt.
 
Last edited:
As long as AMD keeps good price/performance they will steal a big peace of Intel's cake. We will have HEDT from AMD. That is epic enough. And they actually compete with intel.
 
In relation to your complaints about single core performance Bagpuss, really not sure what that is all about. From what i've seen current AMD cpu's are barely 5% behind Intel offerings in terms of IPC.

.err, more like 10-12%, add that to the 20% lower clock speeds and that is a lot of performance you are losing in comparison to a 4.8Ghz clocked Intel CPU.
 
.err, more like 10-12%, add that to the 20% lower clock speeds and that is a lot of performance you are losing in comparison to a 4.8Ghz clocked Intel CPU.

Well actually a 1700X at 3.97GHz up against a 5GHz 7700K isn't as terrible and doomed as you make it out to be, After all the bios/microcode updates it can match and/or beat a 7700K at 5GHz with high speed ram, Not in all games but a good chunk of them.

pOmBzTx.png


2GljI9S.png


sgWbkj1.png


5fdlymN.png


o8wqp6Q.png


XHSM959.png


RqnXLtk.png
 
There were definite errors with the bench RedGamingTech did a video on this leak and it's mentioned in that, I won't post a link but I recommend you see what the issues are for yourself
 
It's total garbage. Yesterday I tested my 8 core Xeon against the score posted here. I got 20,000 overall score with 2600 single thread. That was running all 8 cores at 3.1ghz.

So how this only scores 4000 points more with a massive IPC gain and double the cores with faster HT? god only knows.

I would strongly imagine they ran the benchmark and it basically went "WTF is that?"

It's also not what I would consider to be a proper CPU benchmark either. For most of the test (about 90% of it) my CPU remained at the 1.6ghz base clock without even clocking up. I also saw hardly any load at all on the CPU, even on one core. Only toward the end did it really kick into gear.

Wait for Cinebench results, but also bear in mind that TR is basically two Ryzens stitched together which is basically two quad cores stitched together and that needs support. Apparently AMD have done everything they can to get the latency down so that it acts more like one CPU, but without software support it will be bad.
 
I'm having a hard time believing that leaked benchmark. I just set up my Ryzen 7 1800X build and got my processor running at 4GHz and ran Geekbench: My single core result is higher than a Threadripper? Does not compute.

24JZ26m.jpg
 
Back
Top