Sunday, December 22nd 2019
AMD "Renoir" 4700U Beats i7-1065G7 "Ice Lake" at PCMark 10, 18% Faster than 3700U
AMD's upcoming Ryzen 7 4700U "Renoir" mobile processor posts an 18 percent performance gain in PCMark 10 over its predecessor, the 3700U, according to benchmark results data compiled by Reddit enthusiast _rogame. The 4700U combines a 4-core/8-thread CPU ticking at 2.00 GHz (base) with up to 4.20 GHz boost, according to the SystemInfo module. It also ends up about 2.8 percent faster than a Core i7-1065G7 "Ice Lake" processor.
The 4700U is rumored to feature AMD's new "Zen 2" CPU cores, an iGPU with up to 13 "Vega" NGCUs (832 stream processors), and leverage the 7 nm silicon fabrication process to boost CPU clock-speeds without affecting the ultraportable platform's typical TDP. 3DMark 11 figures of the iGPU surfaced earlier suggesting competitiveness to Intel's Gen11 graphics. "Renoir" also supports the cutting-edge LPDDR4X memory standard, which could further lower platform power draw, while boosting memory clocks as high as 4266 MHz DDR. Below are comparative "best result" screenshots.
Source:
_rogame (Reddit)
The 4700U is rumored to feature AMD's new "Zen 2" CPU cores, an iGPU with up to 13 "Vega" NGCUs (832 stream processors), and leverage the 7 nm silicon fabrication process to boost CPU clock-speeds without affecting the ultraportable platform's typical TDP. 3DMark 11 figures of the iGPU surfaced earlier suggesting competitiveness to Intel's Gen11 graphics. "Renoir" also supports the cutting-edge LPDDR4X memory standard, which could further lower platform power draw, while boosting memory clocks as high as 4266 MHz DDR. Below are comparative "best result" screenshots.
45 Comments on AMD "Renoir" 4700U Beats i7-1065G7 "Ice Lake" at PCMark 10, 18% Faster than 3700U
My bet is that since it will max out at 8C, it will have the two chips placed in line, a bit like Intels mobile CPU's.
- mobile is much harder because of the lower power budget; if AMD has great efficiency on desktop it doesn't mean they can underclock the CPU and get the highest efficiency; things like power circuitry design, idle/sleep speed and implementation, firmware, cooling, uncore power consumption, everything matters; Intel does have a huge advantage on this and I would be extremely surprised if 4700U will have better battery life with same battery size in the same chassis compared to 1065G7.
- regarding iGPU being a monster, no it won't be; it won't be enough for quality gaming; 3780U in surface pro 3 has Vega 11. 4700U has Vega 13, so 2 extra CUs. Nothing major and depending on the memory controller efficiency, it should bring a small increase in GPU horsepower but nothing major;
- regarding CPU performance, while Ice Lake cannot boost to very high frequencies (3.9Ghz max, 4.1Ghz in very short bursts), it does have a definite IPC advantage (10% better IPC compared to Zen 2); this means that 4700U needs to boost at least to 4.3-4.4Ghz to get to parity, but we all know from desktop that 7nm TSMC needs lots of voltage to reach high frequencies and that is very bad for efficiency;
- all in all, I think they stand a very good chance of getting higher up in the mobile space and Intel needs to be on guard and get TigerLake out as soon as possible if they wanna still retain market leadership in mobile. From what we can see in leaks (10% more IPC compared to Ice Lake, 1.5-2X IGP improvement), they stand a very good chance with it.
I understand that space(and cooling) within a laptop is constrained and limited but the physical size of the CPU package is given (by pinout) and for sure can contain 3 chiplets (CCD+CCD+IO) or (CCD+GPU+IO).
Does anyone wonder why this 4700U is a 8c/8t and not a 4c/8t? It has (IMO) direct relation to the 8core (CCD) chiplet design which is universal to all market segments. AMD probably cant "afford" a design change. The chiplet unified achitecture works more than well for any platform and its the key feature that ZEN2 was build upon. Mobile ZEN2 is "delayed" over six months after initial ZEN2 launch I believe due to binning process and TSMCs 7nm delivery capabilities. Still the EPYC chiplets is the high priority for AMD.Cant argue with personal needs... It is what it is. But I get the feeling that you often generalize you personal needs to general needs of the majority of the endusers and for the time being AMD has the better product in almost every case scenario.
So, calling a product pointless before we actually see it in the market and how it does against competition (performance and consumption)... can at least raise some questions and doubts about your personal opinion and why its expressed like yours does.
But Intel already makes a 6C/12T 25W SoC, so AMD went for 8C/8T.
And, of course, 8 cores would be better than 6 cores with SMT... if 25W were enough to provide decent performance.Sure, because I'm the only one generalizing my personal needs to the population. :DOh, please don't go there. No more "every case scenario".
Because at some point I'll ask you about 3 such scenarios and you may say: gaming, rendering and running file compression. :)
Zen2-based Ryzen are good processors - I'm sure most of us agree on that.
Good products? That's not so obvious. It depends what you expect. And you just said I'm generalizing. :)
And just to preempt any possible shortcuts to thinking, I'd like to remind everyone that there is NEVER a circumstance in which the actions of another person, entity or company can be used as an excuse to justify the actions of the company in question...in other words, it doesn't matter if if AMD or another company has murdered someone, it's not a justification for Intel's actions, so don't even try that misdirection. There is no court of law where any defendant on trial for murder can say to the judge: "Your honor, I know I murdered that person, but there are many other people committing murder too, so I should be found innocent." I really can't stand whenever someone points out the underhanded behavior of Intel or Nvidia or any other company and their apologist fanboys attempt to say "...but, but....AMD did so and so..." because it's completely irrelevant and never a justification.
FYI: I'm not an AMD fan, I've just come to dislike Intel and Nvidia (and those are two very different things) and the unrelenting hive-mind of their fanboy armies, so don't even try to shoehorn me into any false binaries that try and label me or anyone else as an "AMD fanboy" just because we dislike Intel and/or Nvidia.
I’m not sure what is your point here...And by this you think that it’s ok to do it too? ...which is something I can accept.
What I cannot accept is that right now you’re trying to make points and statements with frown upon practices and trying to making it legit.So, the word “almost” has lost all of its meaning all of sudden... right!1 out of 2 about this...
1. You didnt understand my point and what I was talking about generalize...
2. You like to play with words with someone that is struggling to put down his thoughts using a foreign language...
which one is it?
Vega is already mature, so it doesn't need any driver? Try that with generic driver Intel UHD :p
Well if you say nothing sort of a feat 8 core 2Ghz base 4.2Ghz turbo beats both 4 core 8 threads with higher 2.3Ghz base 4.9Ghz turbo and 4 core 8 thread with dedicated GTX 1650 Max-Q if you missed it, I'm curious what type of notebook you currently use :rolleyes:
i7-1065G7 is an Ice Lake CPU with 4 cores at a base clock of 1.3 GHz, max boost of 3.9 GHz and all core boost of 3.5 GHz, but this is just burst speed, I believe sustained boost is ~2 GHz. When TDP is configured to 25W, base clock increases to 1.5 GHz.
It's quite impressive that i7-1065G7 does so well with fewer cores and lower (rated) clocks than 4700U. But as I've said, these are rated clock speeds, not actual clock speeds. Also these are synthetic benchmarks.
So while this will make a difference in some use cases, it won't be a differentiating factor between the companies.
The best part is that Tiger Lake should boost PCIe 4.0 adoption by component/accessory makers...
But nothing is confirmed, all are just speculations. It might be duo to the fact AMD can have two options to stay within 15W TDP, it's either 4C8T or 8C8T. Doing the 8C16T route will consume too much power they will have to get the clocks very low to compensate...
I prefer a more general approach, because it makes sense. Because we're talking about products, not how they're made. And products are defined by performance, features, power consumption, size, price etc. They are not defined by architecture, IPC or node size.
People on this forum tend to focus way to much on how CPUs are made and it likely affects their decisions (as consumers, maybe as investors).
The obvious example (on the "investor" side) is the ongoing discussion about AMD's profitability.
Most people on this forum focus on 7nm, chiplets, being fabless, having unified architecture with little waste and so on. It's all amazing, right? Especially in comparison with Intel's large, monolithic dies.
But others say that AMD does not control manufacturing, that they're competing for 7nm supply, that they're fully dependent on a single provider that will exploit this situation.
I can't tell you how many times I've been attacked here for this kind of "heresy". :)
And then the financial statements arrive and guess who was right? :)I'm not sure what this sentence means.
I don't think "Intel-wise". I think "me-wise". If you get an impression that it sounds Intel-ish, that's because it likely does. As I said: I prefer their model and I feel more connected to how they position and market their products. I won't deny that.The point is: you've earlier said
[ICODE]Does anyone wonder why this 4700U is a 8c/8t and not a 4c/8t? It has (IMO) direct relation to the 8core (CCD) chiplet design which is universal to all market segments. AMD probably cant "afford" a design change. [/ICODE]
which sounded like if you though it must be 8 cores because the die design is made for that.
Obviously, AMD could launch a whole lineup of mobile SoCs based on Zen2 die and none of them would go above 4-6 cores.I only believe in how "almost" is defined in math, which is the opposite of how you seem to do it. :P
this IS STILL GOING ON TO THIS DAY... I CAN PROVE IT... i have INTELS x86 Compiler, if i run it on my Ryzen 7 3700X (which has Superior Single thread compared to latest I7/ i9 intel in a friends pc.. the cpu, magically on intels is always faster with "INTELS" x86 Compiler.. seems fishy to me.. you smell the fish too huh, i guess not, but i do, i seen it, i smell it and i hear it too!!!
case closed
"Yes, I prefer Intel's business model and products in general "
seriously , their business model is ANTI-COMPETITIVE in nature... id never support them for that and i never will
I would go to next gen intel or mx250 if I want a cheap ultrabook with some capacity for gaming, amd still got to do their homework in this department. Amd laptops with dedicated gpu are great, ultrabooks look quite bad.