AMD’s "together we advance_PCs”l ivestream presumption conscionable wrapped up moments ago, wherever AMD CEO Dr. Lisa Su acceptable the signifier for the merchandise of the adjacent procreation of AMD Ryzen desktop CPUs. Building disconnected of AMD’s Ryzen 7000 announcement backmost astatine Computex 2022, the eagerly anticipated presumption laid retired AMD’s motorboat plans for their archetypal household of Zen 4 architecture-based CPUs, which volition spot AMD footwear things disconnected with a quartet of enthusiast-focused chips. Topping retired with the 16 halfway Ryzen 9 7950X, AMD’s Ryzen 7000 chips volition beryllium launching successful conscionable implicit 4 weeks’ time, connected September 27^th, with AMD expecting to handily retake the show crown crossed virtually each categories of the PC CPU space, from gaming to contented creation.

Driving AMD’s gains successful this newest procreation of desktop CPUs is simply a operation of architectural improvements underpinning the Zen 4 architecture, arsenic good arsenic moving accumulation of the CPU halfway chiplets to TSMC’s leading-edge 5nm process. The operation of which volition let AMD to present what they are saying is present a 13% summation successful IPC implicit their Zen 3 architecture – up from an 11% assertion arsenic of Computex – arsenic good arsenic a sizable summation successful CPU clockspeeds. The top-end Ryzen 9 7950X volition person a maximum turbo clockspeed of 5.7GHz, 800MHz (16%) higher than the equivalent Ryzen 9 5950X. As a result, AMD expects to present a 29% generational summation successful single-threaded performance, and adjacent much successful multi-threaded workloads.

Launching successful conjunction with the caller Ryzen 7000 bid chips volition beryllium AMD’s AM5 platform, which volition beryllium cornerstone of AMD’s user desktop level done astatine slightest 2025. AM5 introduces DDR5 and PCIe 5.0 enactment for AMD’s chips, arsenic good arsenic a caller LGA socket. 4 chipsets are planned frankincense acold – X670 Extreme, X670, B660 Extreme, and B660 – with X670 bid boards disposable for the September launch, and B660 bid boards acceptable to travel successful October.

Ryzen 7000 Family: 4 Zen 4 SKUs

For regular AnandTech readers and AMD followers, AMD’s merchandise stack for adjacent month’s motorboat should look precise familiar. AMD has done precise good for themselves by starting things disconnected with an enthusiast-focused desktop motorboat that’s built astir a radical of 4 high-end processors, and for the Ryzen 7000 generation, AMD isn’t deviating from this.

As established by AMD backmost astatine Computex, Ryzen 7000 chips apical retired astatine 16 cores. So crossed AMD’s merchandise stack, the halfway counts are the aforesaid successful this procreation versus the last. AMD’s apical SKU volition connection 16 cores, followed by 12, 8, and yet 6 CPU cores. And arsenic before, AMD is gathering their chips utilizing up to 2 Zen 4 CPU halfway chiplets (CCDs), each comprising 8 Zen 4 CPU cores.

As an aside, AMD is not disclosing immoderate further details astir the integrated RDNA2 architecture GPU astatine this time. So portion each of the Ryzen 7000 chips travel with an iGPU, we inactive bash not person specifications to talk of. Regardess, AMD has made it wide astatine aggregate points that the iGPU for these desktop chips is simply a comparatively tiny configuration for basal desktop work, and is not designed to beryllium a high-performance GPU similar connected AMD’s APUs.

Ryzen 9 7950X: The Fastest CPU In the World

Leading the battalion is the Ryzen 9 7950X. The precise champion of AMD’s Zen 4 architecture volition usage 2 afloat enabled CCDs to present a full of 16 CPU cores. Thanks successful ample portion to TSMC’s 5nm process, it comes with immoderate precise precocious clockspeeds arsenic well; the basal clockspeed unsocial is 4.5GHz, and the turbo clockspeed volition scope up to 5.7GHz for a azygous thread. The second is really 200MHz higher than AMD’s Computex presentation, wherever the institution deed just 5.5GHz.

AMD is advertizing this spot arsenic coming with 80MB of cache. That breaks down to 64MB of L3 cache – 32MB connected each CCD – arsenic good arsenic 1MB of L2 cache paired with each Zen 4 CPU core. This is the aforesaid magnitude of L3 cache arsenic recovered connected the Ryzen 5000/Zen 3 family, but the magnitude of L2 cache per halfway has doubled implicit the erstwhile generation.

The 7950X volition besides beryllium AMD’s astir powerfulness bare mainstream desktop Zen spot yet. The TDP for the spot is 170W, and arsenic we cognize from erstwhile AMD disclosures, the Power Package Tracking (PPT) bounds for these chips volition beryllium 230W. This is 1 of the large reasons that AMD is seeing specified important MT show gains – exceeding their mean ST show gains – arsenic erstwhile Ryzen 5000 chips were often TDP constricted erstwhile each of their CPU cores were nether dense load, leaving them good abbreviated of hitting the higher clockspeeds they’d different beryllium capable to support.

Since AMD is the archetypal CPU vendor retired of the gross with this procreation of parts, the 7950X doesn’t person immoderate nonstop contention – though if you’re considering an Intel Core i9 12900K, AMD would beryllium much than blessed to upsell you connected a 7950X instead. According to AMD’s figures the afloat fleshed retired Zen 4 spot should easy beryllium the astir almighty desktop CPU successful the world, thing we’ll of people request to corroborate astatine reappraisal time, but is not hard to judge fixed AMD’s architecture updates and improvements brought by TSMC’s 5nm process. Still, a spot farther down the enactment it volition yet bash conflict with Intel’s Raptor Lake CPUs erstwhile those launch.

AMD is, a small surprisingly, pricing their caller flagship spot astatine $699. This is really $100 beneath the motorboat terms of the 5950X backmost successful precocious 2020, wich deed the streets astatine $799. The higher terms of that spot was not particularly good received, but nary the little for rather a portion AMD sold everything they could make. Still, the spot shortage has abated and AMD has much entree to some wafers and substrates, and the institution volition not beryllium going without contention for the full beingness of the chip.

Ryzen 9 7900X

A measurement beneath the 7950X we person the Ryzen 9 7900X. This is different 2 CCD part, but with lone 6 cores enabled connected each CCD for a full of 12 CPU cores and 24 threads. This portion besides retains the afloat 64MB of L3 cache that comes with the 7950X, which is identical to however the L3 cache scales down (or rather, doesn’t) connected the Ryzen 5000 series.

With its TDP acceptable astatine the aforesaid 170W worth arsenic the 7950X, the 7900X really gains a spot successful basal clockspeeds. This portion volition connection 4.7GHz astatine its base, 200MHz higher than the 7950X. However the maximum turbo clockspeed is 5.6GHz – a spot nether the 7950X – truthful if you privation the implicit champion single-threaded performance, the 7950X is positioned and configured to beryllium the fastest enactment for some ST and MT workloads.

Ryzen 9 7900X volition beryllium priced astatine $549, which is the aforesaid terms arsenic AMD’s erstwhile 12 halfway part, the 5900X. So for a existent apples-to-apples examination connected generational show gains astatine a changeless price, this volition beryllium 1 constituent to look at.

Ryzen 7 7700X

Going farther down AMD’s Ryzen 7000 merchandise stack, we person the Ryzen 7 7700X. This is the 8 halfway Zen 4 counterpart to AMD’s 5800X successful the erstwhile generation. Like its predecessor, AMD is utilizing conscionable a azygous Zen 4 CCD here, but it is simply a fully-enabled CCD. Base clockspeeds volition beryllium 4.5GHz, portion the maximum turbo clockspeed volition apical retired astatine 5.4GHz.

Since it’s conscionable 1 CCD, this besides means that the L3 cache is halved to 32MB. Combined with the 8MB of L2 cache dispersed amongst the Zen 4 CPU cores, and you’ll spot AMD advertizing this spot arsenic offering a full of 40MB of cache.

Meanwhile, dissimilar the 12 and 16 halfway chips, the 7700X volition person a much accepted TDP of 105 watts. This makes the spot easier to powerfulness (and to cool), though it volition beryllium absorbing to spot what this means for turbo clocks nether dense MT workloads. With less CPU cores wide there’s little silicon to airy up, but that means AMD besides isn’t getting the further electrical headroom to propulsion clockspeeds higher.

The 7700X volition deed the thoroughfare with a $399 terms tag, which similar the 7950X, is simply a flimsy terms driblet implicit the erstwhile procreation (where 5800X was priced astatine $449).

Ryzen 5 7600X

Bringing up the bottommost of the Ryzen 7000 motorboat stack is the sole Ryzen 5 people chip, the Ryzen 5 7600X. Like its predecessor this is simply a 6 halfway part, allowing for a full of 12 threads. Driving it is simply a azygous Zen 4 CCD with 2 of the CPU cores disabled.

The basal clockspeed of the spot volition basal astatine 4.7GHz (seeing a akin boost arsenic the 7900X), portion the turbo clockspeed volition beryllium the lowest of each of the archetypal Ryzen 7000 chips astatine 5.3GHz. And yet adjacent arsenic the slowest of the Zen 4 processors, AMD is boldly claiming that the 7600X should, connected average, beryllium 5% faster successful gaming than Intel’s Core i9 12900K (never caput the cheaper chips).

This spot is seeing a TDP bump, however. Whereas AMD’s apical 6 halfway spot successful the erstwhile procreation was 65W, present adjacent the bottommost of the stack is starting astatine 105W. So this should not beryllium considered a petite spot – AMD expects it to some gully and present rather a spot of power.

Zen 4 Architecture Tease: +13% Desktop IPC & AVX-512

Underpinning these caller chips is of people AMD’s Zen 4 architecture. At conscionable fractional an hr long, AMD’s presumption was not astir lengthy capable to dive profoundly into the Zen 4 architecture (AMD has to prevention thing for the launch). But CTO Mark Papermaster did instrumentality the signifier to connection a tease of sorts of further method details connected the Zen 4 architecture.

First and foremost, AMD has revised their IPC estimation for Zen 4. As they person continued to tune the chips and firmware, successful “typical desktop applications” they are present seeing a 13% mean (geomean) summation successful IPC versus Zen 3.

Locking things down astatine an iso-frequency of 4GHz, according to AMD they are seeing anyplace betwixt a 1% summation and a 39% gain, depending connected the workload. The nett effect is that, arsenic is often the case, the nonstop show gains alteration with the task. Still, a batch of AMD’s benchmarks bash travel successful reasonably adjacent to that 13% mark.

For this procreation Papermaster besides brought backmost 1 of our favourite diagrams, the IPC breakdown, showing however the assorted architecture changes to Zen 4 contributed to the wide IPC gain.

According to AMD, the biggest contributors successful this procreation are their caller front-end , followed by load/store improvements, and past an improved subdivision predictor, execution engine, and the larger 1MB of L2 cache per CPU core. Specific method details connected what AMD has changed present volition person to stay for different time, but Papermaster did fto it gaffe that for the execution engine, the operand cache size has been accrued by 50%; truthful it’s wide that AMD has been playing with their assorted caches by rather a spot astatine the CPU halfway level for Zen 4.

Papermaster besides confirmed for the archetypal clip that Zen 4 – including the Ryzen 7000 bid – volition enactment AVX-512 instructions. AVX-512 is simply a spot of a messiness of standards, truthful too the instauration (AVX-512F) instructions, it’s inactive not wholly wide which subsets of AVX-512 AMD volition support. But Papermaster did explicitly notation Vector Neural Network Instructions (VNNI) arsenic among the further subsets supported.

Critically, however, AMD is diverging from Intel successful 1 important aspect: whereas Intel built a true, 512-bit wide SIMD instrumentality for executing AVX-512 instructions, AMD did not. Instead, AMD volition beryllium executing these instructions implicit 2 cycles. This means AMD’s implementation inactive benefits from each of the further instructions, registry record space, and different method improvements that came arsenic portion of AVX-512, but they won’t summation the innate doubling successful SIMD throughput.

In discussing the rationale for AMD’s decision, Papermaster cited the utmost powerfulness requirements for a existent 512-bit SIMD artifact arsenic the biggest impetus for keeping AMD’s SIMD plan astatine 256-bits. As we’ve already seen successful Intel chips with AVX-512 support, the monolithic throughput of a 512-bit SIMD combined with its precocious density results successful a hard spike successful powerfulness depletion erstwhile utilizing it, requiring Intel’s chips to downclock connected AVX-512 workloads (sometimes severely) successful bid to support powerfulness and thermals successful check. Using a narrower 256-bit SIMD means that AMD won’t request to airy up astir arsenic galore transistors astatine once, which volition successful crook marque it easier to support clockspeeds and powerfulness depletion much consistent. At the aforesaid time, I don’t deliberation AMD minds that the dice abstraction requirements for a 256-bit SIMD are importantly little than a 512-bit SIMD; a afloat 512-bit SIMD is simply a batch of transistors to build, and a batch of transistors to occurrence up during dense workloads.

Finally, not to beryllium understated is the interaction TSMC’s 5nm process to AMD’s CPU cores and their resulting chiplets. AMD’s CCD chiplets person utilized TSMC 7nm for some the Ryzen 3000 (Zen 2) and Ryzen 5000 (Zen 3) generations – truthful AMD’s desktop CPUs person been connected 7nm for implicit 3 years now. As a result, the displacement to 5nm has been a agelong clip coming, and the archetypal clip successful rather a portion that AMD has been capable to bask a afloat node shrink for their desktop CPUs.

The benefits of 5nm are evident everyplace from clockspeeds to powerfulness depletion to dice size. The improvements to clockspeeds successful peculiar correspond a large displacement for AMD, whom a mates of generations agone faced a clockspeed shortage versus Intel, and arsenic a effect constricted AMD’s wide single-threaded performance. Now AMD is looking astatine their apical chips going good implicit 5GHz, delivering a meaningful clockspeed boost connected apical of their 13% IPC gains – for a full of a 29% summation successful single-threaded performance.

Meanwhile, contempt AMD’s architectural additions to Zen 4 – including the double-sized 1MB L2 cache – a Zen 4 CPU halfway positive its associated cache is smaller than a Zen 3 halfway and its 512KB of L2 cache. According to Papermaster, the resulting Zen 4 halfway is 18% smaller than its predecessor, measuring 3.84mm² successful size. And, being ever truthful cheeky, helium besides threw successful a examination to the size of Intel’s Golden Cove core, which is recovered connected their Alder Lake processors. According to AMD’s figures, a Zen 4 halfway is outright fractional the size of a Golden Cove core, which gives AMD a important vantage successful dice size (and frankincense astatine immoderate level, manufacturing costs).

We’ve known for immoderate clip that TSMC’s 5nm process has a important pb successful density versus the Intel 7 process, and that is undoubtedly playing a immense portion successful AMD’s resulting size advantage. None the less, arsenic AMD is trying hard to curate the thought of Zen 4 being a leadership-class design, country ratio is 1 much facet that AMD is competing with – and tin clasp implicit Intel’s head.

Finally, successful a passing comment, Papermaster besides disclosed the size of the metallic stack AMD is using. Altogether, AMD is utilizing a 15 furniture telescoping stack.

Zen 4 vs. Zen 3: A Performance Preview

So betwixt their architectural changes and their entree to TSMC’s 5nm process, what does each of this mean for AMD? In short, the institution is rather boldly proclaiming that they person the fastest desktop CPUs connected the planet, and that the Ryzen 7000 chips volition beryllium delivering important improvements implicit some Intel’s existent chips arsenic good arsenic AMD’s previous-generation Ryzen 5000 parts.

And portion each first-party benchmarks should beryllium taken with a atom of salt, present is what AMD is claiming for powerfulness and performance.

We’ll archetypal commencement with iso (like-for-like) comparisons betwixt Ryzen generations, with iso-power.

Interestingly, AMD offered show figures for 3 antithetic TDPs: 65W, 105W, and 170W. The top show gains are really astatine the lowest TDPs, wherever the 7950X saw a 74% summation successful Cinebench R23 MT performance. These advantages really decreased arsenic TDPs went up, dropping to 37% astatine 105W, and yet 35% astatine 170W.

This is meant to underscore however AMD’s absorption connected vigor efficiency, combined with TMSC’s 5nm process, has delivered important gains crossed the board, but particularly astatine little TDPs. When AMD isn’t loading CPU clockspeeds into the stratosphere – which is ever good into the diminishing returns of the voltage/frequency curve – Zen 4 is importantly much powerfulness businesslike than its predecessor. And adjacent erstwhile AMD dose spell for broke connected clockspeeds, Zen 4 inactive comes retired good ahead.

On the whole, AMD is claiming that Zen 4 volition present up to 49% much show than Zen 3 astatine iso-power. Note that this is an “up to” fig and not an average. But it goes to amusement what Zen 4 should beryllium susceptible of erstwhile it gets to enactment its champion ft forward.

Meanwhile, taking things from the different absorption connected Cinebench, AMD besides provided iso-performance figures. According to the company, Zen 4 (7950X) consumes up to 62% little vigor for the aforesaid level of show arsenic a Zen 3 spot (5950X). And portion AMD is going to beryllium plowing astir of those gains backmost into improving show (clockspeeds) for these Ryzen 7000 desktop parts, it’s a further illustration of what AMD is touting to beryllium important ratio gains.

As for however things stack up against arch-rival Intel, AMD is offering likewise bullish figures, albeit with an other grade of cherry picking. In Chaos V-Ray, for example, AMD is claiming that the 7950X offers 47% amended show per watt than the 12900K.

And successful gaming workloads, the institution says that adjacent the (relatively) diminutive 6-core 7600X volition outperform the 12900K by an mean of 5%. Which to beryllium sure, that doesn’t person AMD winning everything, but a 6 halfway configuration is astir arsenic adjacent arsenic you tin get to Zen 4 warring with 1 manus down its proverbial back.

Overall, AMD is claiming that the Ryzen 7000 bid volition present 11% amended gaming show than the 12900K, arsenic good arsenic 44% amended show successful highly multithreaded contented instauration applications. Which, if this proves existent successful autarkic evaluation, surely stands to enactment the Ryzen 7000 good into the pb for this forthcoming procreation of PC processors.

Ryzen 7000: Coming September 27th

Bringing things to a adjacent for now, AMD’s archetypal measurement into the Zen 4 epoch volition footwear disconnected connected September 27^th. That is erstwhile each 4 of the archetypal Ryzen 7000 CPUs volition spell up for sale, arsenic volition the archetypal question of the associated AM5 level motherboards.

According to AMD, portion archetypal motorboat logistics are ever an absorbing situation (especially present arsenic shipping is inactive a small wacky), the wide proviso of Ryzen 7000 chips is expected to beryllium good. While we’re alternatively accustomed to seeing launch-day allocations of caller high-end hardware merchantability out, the wide connection from AMD is that Ryzen 7000 chips should beryllium plentiful. So dissimilar the Ryzen 5000 launch, which was precise adjacent to the highest of the spot and substrate shortage, Ryzen 7000 chips should beryllium much regularly available. And AMD has invested successful the wafer and substrate capableness to assistance guarantee that happens.

Meanwhile connected the motherboard front, the motorboat of AM5 boards volition beryllium dispersed implicit 2 months. For the archetypal September 27^th launch, AMD’s committee partners volition person their high-end X670/X670E boards available. The much mainstream B650 boards – including the recently announced B650E chipset – volition past beryllium a period later. So imaginable buyers volition person to measurement taking an X670 committee astatine launch, versus waiting to drawback cheaper B650 boards.

According to AMD, motherboard prices volition commencement astatine $125. This is presumably for entry-level B650 boards without immoderate PCIe 5.0 functionality. PCIe 5.0 volition undoubtedly bring up the terms connected B650E and different boards, but astatine this constituent we don’t person a bully reply arsenic to conscionable however overmuch PCIe 5.0 volition adhd to the accumulation costs and retail prices of AM5 motherboards.

In conjunction with AM5 boards, the 27^th volition besides spot the merchandise of the archetypal mill overclocked representation kits specifically tuned for the Ryzen 7000 series. For this procreation AMD volition beryllium offering the EXtended Profiles for Overclocking (EXPO) standard, which is AMD’s counterpart to Intel’s XMP DIMMs erstwhile it comes to pre-programed profiles for mill overclocked memory. AMD expects astatine slightest 15 EXPO-enabled representation kits to beryllium disposable astatine launch, astatine speeds up to DDR5-6400.

And with that, the Zen 4 dice is cast. AMD is aiming for thing little than a enactment presumption successful this generation, and travel September 27^th, we’ll get to spot first-hand however good AMD’s engineers tin present connected the company’s ambition. If thing else, aft a pandemic, a spot proviso shortage, and more, it’s going to beryllium bully to get backmost to a much regular CPU motorboat successful the PC space.