The Run

Recently Ryan Shrout demoed a Tiger Lake system running Battlefield 5 at 30 fps on high settings. The demo looked fine, but in the void, it doesn’t mean much, below I leave the original tweet.

Assumptions, assumptions, assumptions

But what were the specs of the system? He didn’t give any info but luckily it’s very easy to guess with the right assumptions, he simply was running a best case configuration, since I’ve been following all of the Tiger Lake leaks I can even give the specs for that!

  • CPU: 1165G7/1185G7 (integrated GPU should be the same perf between both)
  • TDP: 28W
  • RAM: LPDDR4x 4266MHz

Filling the void, how does this compare?

Like I said the run by itself looks good, but the real value of this is how it compares to existing products and I can say this is where things start to get GOOD, first let’s compare with its direct predecessor, Ice Lake, for this I have two sources, first the review of 1065G7 at 1080p by Andreas Schilling:

Second a comparative benchmark by TechEpiphany:

As we can see both are similar and in both cases Tiger Lake doubles(or more) the performance, which if also applies to other games, would mean Intel over delivered on their promise of twice the graphical perf on Tiger Lake over Ice Lake, which was actually made with 15W vs 28W, in this case both ICL systems had more than 15W, so it really does look good, but beware this might be a good scenario for wins since the game previously didn’t do well on Ice Lake.

However, this also has to beat AMD’s Ryzen 4000 mobile series AKA Renoir to truly call it a win, and spoiler alert it does! Similar to the 1065G7 we have two benches done in similar conditions. The first is a 4900HS (8 Vega CUs, 1750 MHz held, 35W, 2xDDR4 3200) which gets 25FPS¹

The second is again TechEpiphany which got an average of about 26 FPS

I must comment that this is a bit strange since TechEpiphany’s system has one less CU and less watts which in this case means lower held clocks, but his result is the same or a bit higher, I don’t know the reason of this but still, Tiger Lake has a significant lead over both, of about 20-25% which is a good result, particularly since Battlefield V tends to do very well on AMD GPUs.

Breaking the “3DMark curse”

As some people have said to me Ice Lake performance on 3DMark was good, however in actual games it was way below, since we have leaks of a perfectly comparable Tiger Lake laptop, we can compare how it does!

This is the comparable Tiger Lake laptop; the graphic score is 1482.

This is the exact same 4900HS that got 25FPS in BFV, the graphic score is 1282, link via @_rogame

And last, we have a 4700U at 25W that is also perfectly comparable to TechEpiphany’s laptop since both are 7CU at 1600 MHz max and 25W power, the graphic score is 1093.

116G7 is 15% faster than the 4900HS and 35% faster than 4700U 25W, the scaling here makes sense unlike the one on the BFV runs, the advantage is also similar to the average of the BFV runs (20-25% for the BFV runs, 25% on time spy). This is a very good hint that the leaked 3DMark perf really is going to be seen in the games, and that’s great news.

Tiger Lake, destroyer of painters, wielder of the power of the plus

To finalize this article I want to say that with this I can confirm that Tiger Lake will be the superior mobile chip, with higher Single/Lightly Threaded performance, higher Graphical performance and better battery life(battery size and laptop implementation being equal) over Ryzen 4000 AKA Renoir.

Also I cant fail to mention how impressed I’m with 10nm++, not only it has surpassed my already high predictions but it has managed far bigger gains than any recent node shrink and has gone from the broken node that was 10nm+ on Ice Lake to outright surpassing TSMC N7 (which is commonly referred as the best node) truly, this is the peak of the #PowerofthePlus!