Intel Core i7-1165G7 ties with the AMD Ryzen 7 4700U in Time Spy CPU Score despite having half the cores while its Xe iGPU batters the Vega 7 and Vega 8 competition

[Otniel Yoreiza]

uhm, but what about the price ? If it cost as high as 4900, then, meh.

not to mention that by the time it hit market, 4700U price will already adjusted

[KennyG]

WOW a whole 15% while costing %50, consuming more power, and months late? LOL get your act together Intel.

Yea and single core performance is gonna be at least 50% faster...
Also, I bet that 95% of people would rather have 50% faster single core than whatever 20% multi core. Also, gaming performance with a dedicated graphics card is gonna destroy AMD

Where the heck did you pull those figures from???
Both APU's have 8 threads,  Time Spy is a very much heavily threaded CPU benchmark. Both CPU's tied, despite the AMD 4700u running at a lower clock-speed and tdp.  GPU is a different story. Intel has that win, no contest.

Because Intel only has 4 cores and still perform the same as AMD 8/8, therefore the single core should perform at least 30% faster because each core has to be faster in order to match it. Also it's only 30% opposed to 50% is because hyperthreading is only around 40% of the performance compared to having actually 8 cores. Also, the Ryzen 7 4700U also is running at 25w along with the Intel one so it draws the same power.

The Intel chip has 4 cores/8 threads and the AMD 4700U has 8 cores/8 threads. AMD could have made a 4 core/8 threads CPU too and it would roughly perform the same.

[Ben Jenkinson]

So what was the memory speed for the AMD laptop? You say what it was for the Intel one. Everyone knows this will effect benchmark scores for both the CPU side and the GPU side of an APU.

[Sds]

With best integrated graphics configuration only in top models, there won't be good budget gaming on iGPU any time soon. Budget I mean 400-600$. From 700$ you can buy 1650 gtx laptop 3500 time spy. So to pay 1000$ for iGPU 1000+ time spy is no good option

[Tov]

Good news for PC industry, this will force AMD to ship Ryzen 5000 out faster and could even make them skip RDNA iGPU for RDNA2.

[arya]

it's a high setting and 1080p. the only reason intel achieved this higher score is just because of AMD's limited 512mb graphic memory. if AMD increases the graphic memory or uses the same speed ram. AMD is still faster.
Loving Intel is like loving a mate who is constantly cheating and raping you.

[_MT_]

Wait, so both CPU's have 8 threads and they tied, despite the AMD running slightly slower clocks and at lower TDP?   The GPU performance is very good news tho. Hopefully it'll kick AMD's Radeon team into gear.

All threads are not made equal. Intel has only 4 cores with two-way SMT (each pair of threads shares resources of a single core). While AMD has 8 cores so the threads don't have to share. It's well known that SMT is inferior to full cores. A single core with two-way SMT (two threads per core) is nowhere near as powerful as two separate cores. How much benefit you get depends on the workload (and implementation). You can actually lose performance. In practical workloads, you're probably looking at +50 % at best. So, a 4C/8T CPU should roughly match 6C/6T CPU. At best. And it could get spanked by a 4C/4T CPU. At worst.

So, why would anyone do it? Adding SMT is much simpler than adding full cores. And it can improve performance. It can also be used quite creatively although I suspect that is more the case in research rather than practice. For example, Xeon Phi has four threads per core.

[rs]

Budget gaming on Intel? That was a good one. Intel won't sell this chip cheap. The problem is, it will be hard to justify premium prices if you can offer only 4 cores. And no, that core comparison was plain dumb. The 4700 has half the cores but the same number of threads. So, unless we don't know utilization of all cores or actual power consumption this says us nothing. AMD's Ryzen 3300X achieves a similar score and has 4C/8T, just like the 1165G7. And I think it wouldn't have been a problem to put it in a 25/35W package if optimized for mobile devices.

Finally Tiger Lake seems to be a step forward for Intel. But I don't see it any better than the older AMD Renoir. Single core performance might be 20% higher or so due to higher clocks and higher IPC. But multicore performance will be way slower at probably still higher power consumption. That small iGPU advantage in synthetic benchmark will likely not translate to real world performance or even turns into a small disadvantage in real games. And don't forget, next year AMD will launch RDNA 2 based mobile chips. Those will offer a completely different level of graphics performance. Even a 12 CU Vega based SoC would kill Xe and that would be easy for AMD to implement. Zen 3 will also significantly increase performance. So, Tiger Lake looks like damage limitation, not more.

[SaltyBiscuits]

I am not a professional.As far as I know,cores don't scale perfectly.For example ryzen 3900x which has an average all cores boost clock of 3.9-4.2ghz should theoretically be nearly twice as fast as ryzen 3600,but the reality is that it scores only 12259 on 3dmark which is just 75% faster than 3600.The same thing goes for gpu cores.4900hs has 8 cus but and 100mhz higher clock speed.It should be 21% faster but it is not.As you can see,4900hs is just 3.6% faster.With some microarchitecture improvements and relatively higher average all cores boost clock (plus things that I mentioned earlier),I don't see why i7 1165g7 couldn't be on par with ryzen 4700u.

[Anthony]

Something not right with those results...

Top AMD APU with an extra 10W TDP has same GPU score?

That doesn't seem right...

[Adriaaaaan]

Or in other words Intel's unreleased next gen GPU beats amds previous gen GPU.  Navi would really best this and rDNA 2 would be a humiliation

[Eeee]

Seriouly? Comparing iGPU with different Ram speed? By the time this Tiger shows up, Zen3 mobile will slap it on its face.

[Gogi]

Threaded cores are around 40% more powerful in benchmarking than the ones without hyperthreading. That means that intel 4c/8t is roughly equivalent to 6c. So the difference between the two CPUs is two cores only. Still, tying up with the AMD chip despite having two less cores is exceptional performance on the  Intel's new mobile chip part

[SaltyBiscuits]

I don't think rdna or even rdna 2  igpus will crush intel gen 12 igpus that easily not wihout more power,more memory bandwidth,new process nodes or something else.It is so much harder than you could imagine to squeeze out more performance at the same power even with new microarchitectures.For example,nvidia rtx 2080 ti has 17.6% lead over gtx 1080 ti in performance per watt at 4k. The extra performance gain actually comes from increasing its power consumption by 18%.From radeon vii to radeon rx 5700xt,the performance per watt lead is reduced to a mere 8.7%.Unless vega igpus in 4700u is heavily underutilised or rdna scales way better than we have expected,we may not see a gigantic leap from gcn to navi.

[Msidragon]

The reason the 4900hs is ahead in the 3dmark is not the core or thread count, its the speed in GHz/ipc. 3dmark is a gaming workload and can only benefit from multicore in the physics test, but the physics test doesn't affect the score by very much since physics in 3dmark uses around 10 threads max, I'm not surprised that the 4 core I7 matched with the 8 thread ryzen 7 since cores and threads behave very similarly in recent years and I think both processors will perform equally well in similar workloads but you can't compare like this because ryzen has a different architecture and behaves differently in certain conditions and Intel generally gets better game optimization from devs but I think this is definitely changing, it would be cool to see Intel and amd swap places in cpu and gpu performance but Intel is still a bit on a high horse for its 10+ year reign in the pricing department in my honest opinion.