AMD Ryzen 5 9600X and Ryzen 7 9700X Review: Zen 5 brings stellar gaming performance

AMD’s six-core 12-thread $280 Ryzen 5 9600X ‘Granite Ridge’ processor arrives packing the brand new Zen 5 structure fabbed on TSMC’s 4nm course of node, promising as much as a 16% enchancment in IPC. The 9600X delivers distinctive worth for avid gamers at its worth level — the chip debuts for $20 lower than its predecessor, however it’s 5% sooner in 1080p gaming than Intel’s $400 Core i7-14700K and 12% sooner than the $300 Core i5-14600K. The eight-core 16-thread Ryzen 7 9700X extends these leads additional.

AMD has additionally made important generational enhancements in productiveness workloads, however Intel’s processors nonetheless maintain the lead in heavily-threaded purposes. The Ryzen 9000X processors produce other benefits, although, delivering well-rounded efficiency inside a mere 65/88W TDP envelope, a 40% lower from the prior-gen fashions. The competing 125/253W Core i5 and i7 seem like energy guzzlers compared. In the end, the Ryzen 9000 collection allows sooner, cooler, and quieter programs with a decrease bar for supporting componentry, thus incomes a spot on our record of the finest CPUs for gaming.

AMD’s Zen 5 launch arrives as Intel struggles to handle long-running instability points that trigger seemingly random BSODs with its Thirteenth- and 14th-Gen processors. Intel’s instability difficulty stems from overvoltage and might doubtlessly influence all its 65W and better CPUs — Ryzen 5 and seven’s direct rivals — with an unknown share of processors experiencing issues. Intel plans to ship a microcode patch to handle the difficulty inside a couple of weeks. It has additionally prolonged its guarantee for the impacted product strains by two extra years. For now, the corporate recommends utilizing conservative energy settings to assist cut back the prospect of errors occurring. These suggestions are mirrored in our testing, successfully giving the Intel chips a haircut in our efficiency benchmarks.

The Ryzen 5 9600X and Ryzen 7 9700X are the primary salvo of AMD’s 4 Zen 5 Ryzen 9000 processors, however a schedule delay resulted within the higher-tier Ryzen 9 9750X and 9900X fashions arriving subsequent week. The brand new Zen 5 chips span from the $280 six-core 12-thread Ryen 5 9600X to the $649 16-core 32-thread Ryzen 9 9950X, successfully protecting the identical segments because the prior-gen Zen 4 Ryzen 7000 fashions. The chips have the identical core counts as their predecessors, with a brand new structure providing important efficiency and effectivity enhancements.

The Ryzen 9000 chips drop into the present AM5 socket, which AMD will assist till 2027+. AMD’s companions will quickly launch (timing varies) new X870/X870E motherboards that may enhance USB 4.0 connectivity and lengthen PCIe 5.0 assist to each chipsets. For now, current X670E, X670, B650E, B650, and A620 motherboards will assist the brand new processors as soon as they’re up to date with new firmware.

AMD cites doubled cache bandwidth as a key enabler of the Ryzen 9000X’s elevated gaming efficiency. Newly added full assist for AVX-512 {hardware} acceleration additionally doubles vector and AI efficiency to offer a giant enhance for productiveness purposes that leverage the directions.

The Ryzen 9000 chips are robust in our take a look at suite, particularly in gaming benchmarks, however they face stiff competitors from AMD’s personal secure — AMD’s current gaming-specialized Ryzen X3D processors proceed to supply the most effective gaming bang in your buck, and the corporate additionally has its next-gen Ryzen 9000X3D processors coming to market quickly, which could give gaming purists pause.

The previous-gen Ryzen 7000 collection processors can be discovered at beneficiant reductions, and that pricing will solely drop as the brand new fashions roll out, so some older fashions would possibly provide a greater worth if the value is true. Intel additionally has but to launch its competing Arrow Lake desktop PC chips, a much-needed reset for the corporate that additionally muddies the waters, although we do not count on to see Arrow Lake chips till late this yr. For now, let’s have a look at how the Ryzen 5 9600X and Ryzen 7 9700X fare in our gaming and productiveness benchmarks.

AMD Ryzen 9000 Specs and Pricing

The Ryzen 5 9600X replaces the Ryzen 5 7600X within the ~$300 worth bracket, however AMD reduce launch-day pricing by $20 in comparison with the prior era. On the pricing entrance, the 9600X contends instantly with Intel’s $300 Core i5-14600K, a lackluster Raptor Lake Refresh member of the family that provides little or no over its predecessor, the Core i5-13600K.

The Ryzen 5 9600X will get a slight 100 MHz clock price enhance to a 5.4 GHz enhance, pushed by the identical Precision Enhance 2 algorithms as earlier than, however the firm additionally dialed the bottom clock again by a whopping 800 MHz, which helps cut back the TDP ranking. The Ryzen 5 7600X got here with a 105/142W TDP/PPT (base/max) ranking, however the transfer to the TSMC 4nm course of and the effectivity of the Zen 5 structure allowed AMD to drastically cut back the 9600X’s ranking to 65/88W, an unbelievable 40% discount. Coupled with the quantity of efficiency the corporate manages to wring from the silicon, AMD says the decrease energy consumption results in a 22% achieve in energy effectivity (perf-per-watt).

The $360 eight-core 16-thread Ryzen 9 9700X will get a 100 MHz bump to a 5.5 GHz enhance, however most different specs stay the identical. Once more, the decrease TDP ranking of 65W can also be a 40W discount in comparison with the prior-gen Ryzen 7 7700X regardless of the same core counts. Base clocks have been decreased by 700 MHz, which clearly helps to rein within the TDP. This chip debuts at $40 lower than the launch pricing of its predecessor and competes with the $380 Core i7-14700K.

The Ryzen 9000 collection runs at a lot decrease temperatures than its predecessors, which helps hold frequency residency excessive for higher efficient frequency and longer enhance durations. This is because of a 15% enchancment in thermal resistance, which AMD says permits the chip to function at a 7C decrease temperature than a Zen 4 chip working on the similar TDP. AMD additionally centered on a extra clever structure of sizzling spots on the die, improved sensor placement, and revamped temperature management algorithms to maintain temperatures in examine.

Like Intel’s competing Okay-series fashions in the identical worth vary, the Ryzen 5 9600X doesn’t have a bundled cooler. Nevertheless, current AM4 and AM5 coolers are appropriate, and the 9600X’s decrease energy envelope ensures that even mid-frame tower air coolers (or equal) present adequate cooling for the processor.

AMD additionally improved the 9600X’s base reminiscence assist from DDR5-5200 to DDR5-5600, however other than an expanded 48KB L1 knowledge cache (L1D) for Zen 5 (extra under), many different key specs stay the identical. AMD says you’ll be able to count on usually larger reminiscence overclocking ceilings that would attain as much as DDR5-8000, with a specific benefit should you go for an 800-series motherboard. That mentioned, DDR5-6000 stays the value/efficiency candy spot for many customers. As soon as once more, AMD helps ECC within the silicon, however implementing that’s as much as the motherboard OEMs.

Ryzen 9000 provides real-time reminiscence overclocking, permitting you to vary reminiscence speeds and timings in real-time from inside the Ryzen Grasp software — no reboot required. Intel additionally has an analogous characteristic. AMD additionally has a brand new Reminiscence Optimized Efficiency Profile that enables guide or automated switching between the JEDEC and EXPO overclocking profiles in actual time to prioritize both bandwidth or latency (timings) for the working software. You possibly can toggle between the completely different profiles within the Ryzen Grasp utility or allow automated profile switching. We’ve not had time to check this characteristic but, however it’s on our to-do record.

AMD has additionally added some new knobs for CPU overclocking. The present Curve Optimizer characteristic lets you set one detrimental or optimistic offset for your entire voltage/frequency curve, which the processor then applies throughout the complete curve. A brand new Curve Shaper characteristic offers rather more granular management by letting you get within the weeds and alter three temperature and 5 frequency factors, leading to 15 management factors.

AMD’s Ryzen Grasp already has an automatic take a look at that assigns the Curve Optimizer offset. That software will proceed to work, however it solely applies a single world optimization setting. Nevertheless, the 2 options can be utilized in tandem, with the consumer assigning extra Curve Shaper factors to the only Curve Optimizer worth. Curve Shaper acts as a kind of world modifier that may alter all of the assigned ranges.

The auto-overclocking Precision Enhance Overdrive (PBO) additionally returns. AMD says Ryzen 9000’s decrease TDP vary allows extra headroom for efficiency positive factors, which we’ll put to the take a look at on the next pages.

AMD 800-series Chipsets

AMD has moved ahead from its 600-series chipsets for Ryzen 7000 to the 800-series for Ryzen 9000, however the brand new chipsets nonetheless leverage the AM5 LGA1718 socket. The 800-series motherboards received’t arrive till later within the yr, with some fashions coming in late September and others following primarily based on numerous OEM schedules. The Ryzen 9000 chips are totally appropriate with current AM5 motherboards (after the requisite BIOS replace), so there are many choices for early adopters. The staggered launch is clearly not excellent, however at the very least assist for future chips will likely be round for the lengthy haul — AMD plans to assist the AM5 socket till 2027+.

The upper-tier X870 and X870E chipsets include welcome enhancements, however these advances stem from new obligatory characteristic necessities as a substitute of expanded chip connectivity (extra on that under). As an illustration, the PCIe 5.0 interface is now customary on the X870 lineup for each storage and graphics, whereas it was beforehand restricted to the E-series boards. All X870 boards will now even have the USB 4.0 60 Gbps interfaces courtesy of third-party controllers, just like the ASMedia ASM4242 (the controller will eat a number of the PCIe lanes from the CPU to function at full bandwidth). The 800-series is constructed across the similar Promontory 21 chipset silicon from ASMedia because the 600-series.

The high-end boards are good, however Ryzen 5 9600X is a pure pairing for a reliable B850 motherboard. AMD’s B850 collection options largely the identical characteristic set as the present lineup of B650 motherboards, with full assist for overclocking each the processor and reminiscence and assist for USB 3.2 20 Gbps. AMD hasn’t but revealed the complete breakdown of the ports, however other than non-obligatory USB 4 assist, we’ve gleaned that the B-series characteristic set stays the identical.

Along with the lower-tier B850 boards we anticipated, AMD has added a brand new tier for the B-series with the brand new B840 collection. This chipset bridges the hole between the A- and B-series to assist cut back platform prices. The B840 collection helps reminiscence overclocking, however CPU overclocking isn’t supported. They may also include solely the PCIe 3.0 interface, a giant discount from the combination of PCIe 5.0 and 4.0 interfaces on the B850 household. The B850 and B840 additionally get one other value discount by way of assist for USB 3.2 as a substitute of the obligatory USB 4 on the X-series motherboards. We don’t know the way a lot the decreased characteristic set will influence pricing, however the B840 lineup appears finest fitted to OEMs and workplace machines.

AMD Granite Ridge SOC

AMD’s Ryzen 9000 household makes use of the identical I/O Die (IOD) because the Ryzen 7000 chips, and the general SoC has the identical normal structure. As a result of the IOD stays the identical, the Ryzen 9000 CPUs have the identical assist for 28 lanes of PCIe 5.0, 5 USB ports, and 4 show streams from the built-in RDNA 2 graphics engine. The RDNA 2 iGPU has no practical adjustments, so efficiency needs to be practically equivalent to Ryzen 7000 — the engine is supposed to mild up a show and never rather more.

The IOD, which serves because the communication hub for the processor, is paired with both one or two eight-core CCDs (Core Chiplet Die) — the 9600X has a single CCD, whereas the Ryzen 9 fashions have two CCDs. Processors with a single CCD have a 32B/cycle learn/write port for communication to the IOD by way of an Infinity Material connection. Nevertheless, as earlier than, dual-CCD chips have a 16B/cycle write and 32B/cycle learn connection between the IODs to save lots of energy on the high-power SERDES and densify the package deal structure.

The Granite Ridge ‘Eldora’ CCD has 8.6 billion TSMC N4P transistors unfold throughout 70.6mm^2 of silicon, equating to a transistor density of 121.81 million transistors per sq. millimeter (MTr/mm^2) — a 31% enhance in transistor density over Zen 4’s Durango CCD.

As earlier than, the IOD has 3.4 billion TSMC 6nm transistors unfold throughout 122mm^2 of silicon. As such, a single-CCD Ryzen 5 9600X has a complete of 12 billion transistors, whereas the dual-CCD Ryzen 9 fashions have 20.6 billion.

The TSMC N4P (4nm) node

Zen 5 processors use TSMC’s N4P node, a marked enchancment over Zen 4’s 5nm node, however AMD can have 3nm variants coming to market sooner or later. TSMC’s N4P delivers a number of the finest generational enhancements of the corporate’s 4nm choices (TSMC’s 4nm node is formally a part of its 5nm household).

TSMC says the N4P node provides 11% extra efficiency, 22% larger energy effectivity, and 6% larger transistor density (optical shrink) than the N5 (5nm) node. Moreover, as a consequence of an elevated variety of EUV layers, this course of makes use of 6% fewer masks for manufacturing, which usually equates to being cheaper. The N4P course of node clearly helps ship sooner and extra power-efficient processor efficiency.

The Zen 5 Microarchitecture and 16% IPC enchancment

We’ve written extensively in regards to the Zen 5 microarchitecture, and you will discover extra particulars in our Zen 5 microarchitecture deep-dive, our article with extra Zen 5 particulars, and our interview with AMD’s ‘Father of Zen,’ lead architect Mike Clark.

The Zen structure debuted with a 52% IPC enchancment over Bulldozer in 2017, and AMD has delivered double-digit share IPC will increase with each era since. In actual fact, AMD claims it has delivered a 4.1x complete enhance in IPC because the Piledriver period (primarily based on single-thread Geekbench 5 outcomes).

Based on AMD’s Zen 5 benchmarks throughout 13 workloads, Zen 5 has a 16% enhance in IPC. Zen 5 has dramatically improved vector math efficiency, with a 32% achieve over Zen 4 in single-core machine studying (VNNI) and a 35% achieve in single-core AES-XTS encryption workloads (AVX-512), as measured by AMD’s Geekbench subtests.

We additionally ran our personal IPC assessments, however I found an inconsistency in our outcomes, so I am retesting to right the error. We’ll add our IPC outcomes to this part quickly.

The highlights of the Zen 5 structure embrace a redesigned entrance finish with improved fetch, decode, and dispatch parts that feed a wider execution engine. Zen 5 additionally doubles the info bandwidth between its L2 and L1 caches and the L1 to floating level unit over Zen 4, which AMD says helps ship the generational gaming efficiency enhancements.

Zen 4 employed a dual-issue AVX-512 pipeline (‘double-pumped’ in AMD lingo), that means it issued an AVX-256 instruction twice throughout a 256-bit interface to offer many of the efficiency advantages of AVX-512 whereas avoiding giant impacts to die space and frequency drops.

Zen 5 helps the complete 512-bit knowledge path to ship doubled AVX-512 and VNNI throughput, however AMD says the chip nonetheless runs on the similar frequency as integer work for any given multi-core workload throughout AVX-512, a byproduct of its cautious balancing act with the facility traits of the integer instruction path. That stands in stark distinction to Intel’s AVX implementation, which leads to frequency reductions.

Different normal enhancements embrace a dual-pipe fetch and higher department prediction accuracy, dual-ported instruction and operation caches, a twin four-wide decode path, and an eight-wide dispatch. Zen 4 had a 32KB knowledge cache, which has now been expanded to a 48KB 12-way L1 knowledge cache (L1D) for Zen 5.

AMD outlined the assorted contributors to its 16% IPC enchancment. The transfer to 8-wide execution/retire pipelines contributed 34% of the achieve, 27% every from the decode/opcache and knowledge bandwidth enhancements, and the remaining 12% from the improved fetch/department prediction.

Now, let’s have a look at what these enhancements seem like in motion on the next pages.