Ryzen 7000 I/O Die: TSMC & Integrated Graphics at Last

We’ll start our look at the Ryzen 7000 silicon with the glue that holds the chip together: the I/O Die (IOD). Whether it’s a 7600X or 7950X – one CPU chiplet or two – every Ryzen 7000 processor gets one of AMD’s new IODs.

As with the Ryzen 3000/5000 series, the IOD is a die separate from the CPU core chiplets that hosts everything that isn’t the CPU cores or their associated L1/L2/L3 caches. By producing a 2 or 3 chiplet-based solution rather than a single monolithic die, AMD is able to put silicon that doesn’t require cutting-edge performance on a cheaper process node, while enjoying the yield advantages that come from smaller chiplets and older, more mature nodes.

And yet with that said, for the Ryzen 7000 series it’s AMD’s IOD that’s getting the biggest upgrade. Previously produced on GlobalFoundries’ 14nm/12nm process node, for Ryzen 7000 the IOD has made the leap to TSMC. The new IOD is made on TSMC’s 6nm process, a generation behind the 5nm node used for the CPU chiplets themselves, but a generation (and then-some) ahead of GlobalFoundries’ smallest node.

With AMD no longer tightly bound to GloFo with wafer share agreement exclusivity requirements, AMD is now free to produce chips like the IOD where they see fit. And with AMD also doing very well as a company in the last few years, they do not need to produce IODs at GloFo to meet their purchase requirements with the fab.

The jump to TSMC 6nm brings several advantages to AMD’s central chiplet. The much smaller process node allows for significantly more transistors – 3.4B on a 122m2 die, 63% more than before – which AMD can invest in new features such as integrated graphics. As well, the more efficient process node brings down the power requirements of the IOD, especially in near-idle situations, though AMD is relying on a lot more than process nodes here.

In terms of functionality, the IOD first and foremost is responsible for the CPU-side implementation of all the new features of the AM5 platform. So PCIe 5.0 root complexes, DDR5 memory support, updated display I/O, and more are implemented here. Relative to the Ryzen 3000/5000 IOD then, there is almost nothing here that AMD hasn’t touched in some function; even the USB controllers and infinity fabric have been altered for this generation of processors.

Ryzen 6000 Mobile’s Legacy: Power Control and Pluton

Outside of the obvious gains from moving to a smaller process node, the Ryzen 7000 IOD has also given AMD a chance to implement everything they’ve learned in the past 3 years about power control and power optimizations from the Ryzen mobile lineup. Since the previous-generation IOD was launched for the Ryzen 3000 family in 2019, AMD has launched the Ryzen 6000 Mobile CPU, a monolithic die design that went all-out on power efficiency improvements in a bid to catch up to Intel. Ryzen 6000 Mobile introduced power features such as SVI3, which have since been ported to the Ryzen 7000 design, and principally show up as parts of the IOD.

This was especially easy for AMD as the IOD is made on the same 6nm process as the Ryzen 6000 Mobile chips. Which means that the new IOD has been able to implement many of the same power gating, power binning, deep sleep modes, clock gating, and power state features that were first introduced for mobile. Which is to say that AMD has put a lot of hard work into keeping the IOD from working hard.

The Ryzen 7000 also inherits many of the basic security features of the Ryzen 6000 Mobile platform. AMD is still using their own Arm-based security processor within the IOD. And the new chip is compliant with Microsoft’s Pluton initiative as well – with all the mixed responses that will undoubtedly come from that.

Ryzen 7000 Becomes Whole: Integrated Graphics At Last

Of the additional 1.31B transistors in AMD’s new IOD, the single biggest spend against that budget comes from the long-awaited inclusion of integrated graphics on high-end Ryzen desktop parts. Up to this point, AMD’s high-end consumer Ryzen processors have not included integrated graphics; both the early monolithic parts and later chiplet-style parts have forgone them for cost and integration reasons. But now with AMD’s modern RDNA 2 graphics architecture and TSMC fabrication process, AMD has finally seen the (ray traced) light, and is building a small GPU block into the IOD to offer integrated graphics throughout the Ryzen 7000 product stack.

To be sure, this iGPU is not a very powerful one – and it’s not meant to be. With just 2 CUs (technically a single RDNA 2 dual compute unit), it offers one-sixth the number of CUs on AMD’s top-end Ryzen Mobile APUs, or in the bottom-of-the-stack discrete Radeon RX 6400. So this is not a configuration meant to replace AMD’s traditional APUs in laptop or desktop scenarios. None the less, it is a full implementation of the RDNA 2 architecture – ray tracing and all – in a very low performance configuration.

AMD GPU Specificaiton Comparison
  AMD Radeon Graphics
(Raphael)
AMD Radeon 680M
(Rembrandt)
AMD Radeon RX 6400
Stream Processors 128
(2 CUs)
768
(12 CUs)
768
(12 CUs)
Boost Clock 2200MHz 2400MHz 2321MHz
Peak Throughput (FP32) 0.56 TFLOPS 3.7 TFLOPS 3.6 TFLOPS
Memory Clock DDR5-5200 LPDDR5-6400 16 Gbps GDDR6
Memory Bus Width 128-bit 128-bit 64-bit
Infinity Cache N/A N/A 16MB
Manufacturing Process TSMC N6 TSMC N6 TSMC N6
Transistor Count N/A N/A 5.4B
Architecture RDNA2 RDNA2 RDNA2
GPU Raphael Integrated Rembrandt Integrated Navi 24

The idea behind including an iGPU here is to offer a basic implementation of modern graphics across the entire Ryzen 7000 lineup. Even though it’s not very fast, 2 CUs running at up to 2.2GHz is going to be fast enough for desktop work – and maybe even some very light gaming. This makes the high-end Ryzen chips suitable for desktop work and inclusion in OEM desktop systems without a discrete GPU, a market segment that AMD was previously locked out of and Intel enjoyed free reign. And even for users who are bringing their own GPU, an integrated GPU makes troubleshooting all the easier, as now it doesn’t require a separate video card to merely boot a system.

Besides enabling basic pixel processing, the other major function of the Ryzen 7000 iGPU is to enable all of the video decoding and video output functionality expected from a modern GPU. As this part is derived from the integrated GPU that when into Ryzen 6000 Mobile (Rembrandt), it comes with a very familiar feature set. That includes AV1, HEVC, and H.264 video decoding, as well as HEVC and H.264 video encoding.  And, as AMD is keen to point out, these video encode/decode blocks remain accessible even with a discrete GPU in play; so unless a user outright disables the iGPU, every Ryzen 7000 system will have access to a modern suite of video encode and decode features.

Meanwhile on the display controller side of matters, this is the block that’s enabling Ryzen 7000 CPUs to drive up to 4 4K@60Hz displays. The iGPU display controllers can drive HDMI 2.1 up to its maximum 48Gbps data rate, or it can drive a DisplayPort 2.0 output at up to the UHBR10 data rate (a feature not even found on NVIDIA’s forthcoming RTX 40 series cards). And as mentioned previously when talking about motherboards and chipsets, motherboard vendors will have the option of exposing these DP outputs either via USB-C alt mode, or by implementing fixed DisplayPorts. All of which, in turn, can be used as active display outputs even if a discrete video card is installed, via AMD’s new hybrid graphics mode.

All told, AMD is taking a similar tack as Intel here with their desktop CPUs. Most (but not all) Intel desktop parts contain a basic UHD-class integrated Xe-LP GPU, which Intel does for many of the same reasons as AMD now is. AMD has a leg-up on Intel in as much as they offer a newer architecture and features to much, such as AV1 decoding. But we’ll have to see where performance lands; 2 Radeon CUs is low enough that these parts may not even catch up to Intel’s 32 EU integrated graphics.

AMD Infinity Fabric: Faster Flowing, Narrower Going

The last major functional update to come with AMD new IOD is the Infinity Fabric, AMD’s die-to-die interconnect. On the chiplet-based Ryzen desktop parts, IF is used to provide a low-latency connection between the IOD and the CPU core chiplet(s) on the package.

For the Ryzen 7000 series, AMD has done some further engineering work to reduce the power costs of the IF links, and thus improve the overall power efficiency of the link and reduce the data movement power penalty for using a chiplet-based approach. The most significant thing here is that AMD has doubled the IF link frequency, but cut the physical width of the IF link in half. As a result, Ryzen 7000’s IF links offer the same bandwidth as on Ryzen 5000,

According to AMD, this refactoring is because AMD’s engineers found that, contrary to conventional wisdom, they could save power running a narrower IF link at a higher frequency. Which is to say that, even though higher frequencies are normally power prohibitive, in the case of IF it’s just powering the physical links themselves which is the most expensive part. Part of that shift, in turn, comes from the fact that the newer process nodes have allowed AMD to run the IF link at a lower voltage,

Finally, further reducing AMD’s power consumption here, the IF links on the Ryzen 7000 now support multiple power states, versus the binary on/off states of the IF links on Ryzen 5000 and earlier. This gives AMD a nice middle ground to save power in lighter load situations, by being able to use intermediate states rather than running at full-power just to have an IF link active.

And while this leaves AMD with an obvious (if more power-hungry) route to take for more IF bandwidth in the future should they need it, for now the company intends to stay pat. According to AMD’s engineers, the amount of bandwidth supplied by the Ryzen 7000 IF links, though virtually identical to the amount of bandwidth on the Ryzen 5000 links, is still enough to meet AMD’s die-to-die bandwidth needs – even with extra memory bandwidth afforded by the switch to DDR5.

DDR5 & AMD EXPO Memory: Memory Overclocking, AMD's Way Zen 4 Architecture: Power Efficiency, Performance, & New Instructions
POST A COMMENT

205 Comments

View All Comments

  • Tomatotech - Friday, September 30, 2022 - link

    Nice idea but you’re swimming against the flow of history. The trend is always to more tightly integrate various components into smaller and smaller packages. Apple have moved to onboard RAM in the same package as the CPU which has bought significant bandwidth advantages and seems to have boosted iGPU to the level of low-end dGPUs.

    The main takeaway from your metaphor of the 650w dGPU with a 55w mainboard and 100-200w CPU is that high-end dGPUs are now effectively separate computers in their own right - especially as a decent one can be well over 50% of the cost of the whole PC - and are being constrained by having to fit into the PC in terms of physical space, power supply capacity, and cooling capacity.

    It’s a shrinking market on both the low end and high end for home use of dGPU, given these innovations and constraints and I don’t know where it’s going to go from here.

    Since I got optic fibre, I’ve started renting cloud based high-end dGPU and it has been amazing albeit the software interface has been frustrating at times. With symmetric gigabit service and 1-3ms ping, it’s like having it under my desk. I worked out that for unlimited hours and given the cost of electricity, it would take 10 years for my cloud rental costs to match the cost of buying and running a home high end dGPU.

    Not everyone has optic fibre of course but globally it’s rolling out year by year so the trend is clear again.
    Reply
  • Castillan - Wednesday, September 28, 2022 - link

    "

    clang version 10.0.0
    clang version 7.0.1 (ssh://git@github.com/flang-compiler/flang-driver.git
    24bd54da5c41af04838bbe7b68f830840d47fc03)

    -Ofast -fomit-frame-pointer
    -march=x86-64
    -mtune=core-avx2
    -mfma -mavx -mavx2
    "

    ...and then later the article says:

    "The performance increase can be explained by a number of variables, including the switch from DDR4 to DDR5 memory, a large increase in clock speed, as well as the inclusion of the AVX-512 instruction set, albeit using two 256-bit pumps."

    The problem here being that those arguments to Clang will NOT enable AVX-512. Only AVX2 will be enabled. I verified this on an AVX512 system.

    To enable AVX512, at least at the most basic level, you'll want to use "-mavx512f ". There's also a whole stack of other AVX512 capabilities, which are enabled with "-mavx512dq -mavx512bw -mavx512vbmi -mavx512vbmi2 -mavx512vl" but some may not be supported. It won't hurt to include those on the command line though, until you try to compile something that makes use of those specific features, and then you'll see a failure if the platform doesn't support those extensions.
    Reply
  • Ryan Smith - Friday, September 30, 2022 - link

    Correct. AVX-512 is not in play here. That is an error in analysis on our part. Thanks! Reply
  • pman6 - Thursday, September 29, 2022 - link

    intel supports 8k60 AV1 decode.

    Does ryzen 7000 support 8k60 ??
    Reply
  • GeoffreyA - Monday, October 3, 2022 - link

    The Radeon Technology Group is getting 16K ready. Reply
  • yhselp - Thursday, September 29, 2022 - link

    I'd love to see you investigate memory scaling on the Zen 4 core. Reply
  • Myrandex - Thursday, September 29, 2022 - link

    The table on page four mentions "Quad Channel (128-bit bus)" for memory support. Does that mean we could have a 4 memory slot solution, with one memory module per channel, with four channel support? This way to drastically increase memory bandwidth all while maintaining those fast DDR5 frequencies? Reply
  • Ryan Smith - Friday, September 30, 2022 - link

    No. That configuration would be no different than a 2 DIMM setup in terms of bandwidth or capacity. Slotted memory is all configured DIMMs; as in Dual Inline Memory Module. Reply
  • GeoffreyA - Friday, September 30, 2022 - link

    All in all, excellent work, AMD, on the 7950X. Undoubtedly shocking performance. Even that dubious AVX-512 benchmark where Intel used to win, Zen 4 has taken command of it. However, lower your prices, AMD, and don't be so greedy. Little by little, you are becoming Intel. Don't be evil.

    Thanks, Ryan and Gavin, for the review and all the hard work. Much appreciated. Have a great week.
    Reply
  • Footman36 - Friday, September 30, 2022 - link

    Yawn. I really don't see what the big fuss is about. I currently run 5600X and was interested to see how the 7600X compared and while it does look like a true uplift in performance over the 5600X, I would have to factor in cost of new motherboard and DDR5 ram! On top of that, the comparison is not exactly apples to apples in the testing. 7600X has a turbo speed of 5.3, 5600X 4.6. 7600X runs with 5200 DDR5 and 5600X 3200 DDR4, 7600X has TDP 105W, 5600X 65W. If you take a look at the final page where the 7950X is tested in ECO mode which effectively supplies 65W instead of 105W you lose 18% performance. If we try to do apples to apples and use eco mode with 7600X, to get apples to apples with 65W of 5600W, then lower boost to 4.6ghz then the performance of the 2 cpu's looks very similar. Perhaps not the way I should be analyzing the results, but just my observation.... Reply

Log in

Don't have an account? Sign up now