An early 28-core Intel desktop chip just posted a single-thread score of 5,210 in PassMark, and the number is pulling attention across the PC hardware press. TechPowerUp reported the leaked Nova Lake-S result on October 28, 2026, and VideoCardz covered the same database entries. The sample also logged an overall CPU Mark of 61,509 and a reported maximum boost clock of 4.8 GHz.
The headline is the single-thread figure. It lands ahead of Intel’s own Core Ultra 9 285K (5,059), AMD’s Ryzen 9 9950X3D (4,805) and the older Core i9-14900KF (4,635). The multi-thread picture is less flattering. The same 28-core part trails both current flagships in overall CPU Mark, which is a hint about how Intel may be tuning this early silicon.
Every number here comes with a large asterisk. The entries sit under a placeholder label, we found no official Intel confirmation, and no retail specification exists. If you are weighing an upgrade, our AM4 vs AM5 breakdown of the Ryzen 7 5800X3D and 9850X3D covers what you can buy today, while this story is about what might arrive next.
Don't miss new tech stories on Google
Add Tech Insider once in the Google app and our stories appear in your news suggestions.
What the Nova Lake-S Leak Actually Shows
According to TechPowerUp and VideoCardz, PassMark’s database now holds entries for a 28-core processor tied to Intel’s Nova Lake-S desktop family. The best run shows 5,210 points in the single-thread test and 61,509 points in the overall CPU Mark. A second entry scored 5,191 single-thread and 60,964 overall. Both used PassMark PerformanceTest 11.1.
The reported maximum boost frequency for the sample is 4.8 GHz. That detail matters because it frames how much of the single-thread result comes from clock speed and how much comes from the core design. We dig into that below.
Several outlets ran with the story. TechSpot framed it as a 28-core Nova Lake CPU beating the Core Ultra 9 285K and the Ryzen 9 9950X3D in single-thread testing. Club386 led with the opposite half of the same data, noting that the chip falls behind in multi-thread work. Notebookcheck called the single-thread result promising for an early benchmark. All three readings are accurate, and together they describe a chip that looks strong per core and unfinished as a whole.
How the PassMark Entries Surfaced
The results were attributed to database entries labeled “Intel 0000.” That is a placeholder string, not a marketing name, and the entries are not official Intel benchmark disclosures. PassMark’s database accepts results from anyone who runs the software, so pre-release hardware in the hands of partners, motherboard makers or testers can appear before the manufacturer says a word.
Why the placeholder label matters
When a retail chip identifies itself, the database shows a clean product name. When firmware has not finalized the brand string, it reports a generic vendor label instead. The “Intel 0000” tag therefore tells you the sample ran on pre-release firmware or an engineering platform. Reviewers who watch these databases treat such entries as directional hints and nothing more.
Who matched the entries to Nova Lake-S
The link to the Nova Lake-S family is an identification made by the reporting outlets, not a statement from Intel. The same goes for the model name Core Ultra 9 4970K BFC, which reports offer as the likely identity of the 28-core sample. We treat both as unconfirmed throughout this article.
Single-Thread Scores Against Intel and AMD Rivals
Single-thread performance drives responsiveness in everyday software, most games and a long list of lightly threaded workloads. On that measure the leaked sample looks good. The table below uses the scores reported by TechPowerUp and VideoCardz, with the percentage gaps calculated from those figures.
| Processor | PassMark single-thread | Nova Lake-S lead (5,210 run) | Lead (5,191 run) |
|---|---|---|---|
| Nova Lake-S 28-core, best run (reported) | 5,210 | n/a | n/a |
| Nova Lake-S 28-core, second run (reported) | 5,191 | n/a | n/a |
| Intel Core Ultra 9 285K | 5,059 | +151 (+3.0%) | +132 (+2.6%) |
| AMD Ryzen 9 9950X3D | 4,805 | +405 (+8.4%) | +386 (+8.0%) |
| Intel Core i9-14900KF | 4,635 | +575 (+12.4%) | +556 (+12.0%) |
A 3.0% edge over the Core Ultra 9 285K is real but modest. A 12.4% edge over the Core i9-14900KF is more meaningful, since that chip is still a common upgrade starting point for builders. The 8.4% lead over the Ryzen 9 9950X3D is the figure rival-watchers will quote, though a PassMark single-thread number says little about how AMD’s cache-heavy parts behave in games.
One more detail stands out. The two Nova Lake-S runs sit only 19 points apart, a spread of roughly 0.4%. That tight grouping suggests the benchmark ran cleanly on stable hardware, which lends the single-thread claim more weight than a lone outlier would.
Multi-Thread Results: Where 28 Cores Fall Short
Core count suggests the 28-core sample should dominate in overall throughput. It does not. The best run reached 61,509 CPU Mark, and the second reached 60,964. Both sit below the current flagships from each rival, according to the comparison scores in the reports.
| Processor | PassMark CPU Mark | Gap vs 61,509 run | Gap vs 60,964 run |
|---|---|---|---|
| AMD Ryzen 9 9950X3D | 72,066 | Nova Lake-S trails by 10,557 (14.6%) | Trails by 11,102 (15.4%) |
| Intel Core Ultra 9 285K | 67,274 | Trails by 5,765 (8.6%) | Trails by 6,310 (9.4%) |
| Nova Lake-S 28-core, best run (reported) | 61,509 | n/a | +545 (+0.9%) |
| Nova Lake-S 28-core, second run (reported) | 60,964 | -545 (-0.9%) | n/a |
A 28-core chip that scores 8.6% below the Core Ultra 9 285K points to an unfinished configuration. Early samples often ship with conservative power limits, immature firmware and unoptimized memory settings. All three drag down sustained multi-core runs far more than they hurt a short single-thread test.
Club386 drew the same conclusion in its headline, pairing the single-thread win with a multi-thread shortfall. We would add that PassMark’s CPU Mark blends many sub-tests, and some of them respond to memory latency and cache behavior as much as to raw core count. A platform still in bring-up will rarely hit its ceiling there.
Reading the Numbers Per Clock
With a reported 4.8 GHz boost and a 5,210 single-thread score, the sample delivers about 1,085 PassMark points per GHz. That is simple arithmetic from the two reported figures, and it is a rough yardstick, since real chips do not hold their peak boost for the entire test. Still, it hints that the core design carries the result rather than a clock speed advantage.
That reading also explains why enthusiasts reacted warmly. If retail firmware raises the boost ceiling, the single-thread number has room to climb. If retail parts clock lower to hold power in check across 28 cores, the score could slip. Nobody outside Intel and its partners knows which direction the final tuning will take.
Treat the per-GHz figure as a talking point for forums and not as a forecast. Maximum boost in a database listing is also a reported value, and the fact sheet we work from does not confirm base clocks, cache sizes or power limits.
Why “Intel 0000” Entries Demand Caution
Engineering samples differ from retail chips in ways that move benchmark scores in both directions. Microcode changes between steppings. Firmware updates alter boost behavior. Test boards used internally often run with debug features that cost performance. A retail Nova Lake-S chip could score higher, lower or about the same as the leaked entries.
There is also the question of the benchmark itself. PassMark is a synthetic suite. It does not run the games, compilers, render engines or AI inference workloads that decide purchases. A 3% single-thread edge in PassMark rarely translates into a visible 3% edge in real software. Independent reviews with retail hardware, final firmware and matched memory remain the only reliable test.
Finally, database entries can be edited, mislabeled or even faked. The outlets matched the figures to Nova Lake-S on their own judgment, and nothing in a public listing can prove a result came from the claimed silicon.
What Is Confirmed and What Is Not
The gap between reported and verified information is wide here, so it helps to separate the two. The table below lays out what the sources support and what remains an open claim.
| Item | Status | Basis |
|---|---|---|
| 28-core sample appears in PassMark | Reported | TechPowerUp, VideoCardz |
| Single-thread 5,210 and 5,191 | Reported | Database entries labeled “Intel 0000” |
| CPU Mark 61,509 and 60,964 | Reported | Same database entries |
| 4.8 GHz maximum boost | Reported | Listing data cited by outlets |
| Identity as Core Ultra 9 4970K BFC | Unconfirmed | Outlet identification, no Intel statement |
| Q1 2027 launch window | Unconfirmed | Claim mentioned in reports |
| Socket, cache and retail specs | Unconfirmed | No official disclosure |
| Price | Unknown | None found |
| Final retail performance | Unconfirmed | Sample may differ from shipping parts |
We found no official Intel confirmation of the benchmark entries and no on-record statement from the company or from any named engineer about the sample. That is why this article paraphrases outlets and avoids attributing opinions to individuals.
Core Ultra 9 4970K BFC: A Name Without Confirmation
The tag Core Ultra 9 4970K BFC comes from reports that match the 28-core sample to a likely retail identity. Intel has not announced a product by that name. The suffix, the numbering scheme and even the “K” designation could all change before launch.
Naming leaks are notoriously unreliable at this stage, because brand strings are often among the last items finalized. We include the name because readers will search for it, and because it appears across multiple reports. You should read it as a working label, not a product page.
The same caution applies to the socket and cache configuration, which no source has confirmed. Anyone planning a motherboard purchase around this chip is guessing. Nothing in the reports establishes platform compatibility with current boards.
Intel Versus AMD at the Top of the Desktop Stack
The comparison scores reveal a split market. Per the reported figures, Intel’s Core Ultra 9 285K leads AMD’s Ryzen 9 9950X3D in single-thread PassMark by about 5.3% (5,059 against 4,805), while AMD leads Intel in CPU Mark by about 7.1% (72,066 against 67,274). Each company holds one half of the crown.
The leaked sample suggests Intel’s next desktop generation wants to widen the single-thread lead while closing the multi-thread gap through core count. On these early numbers it pushes the first goal forward and has not yet reached the second. If Intel’s engineers lift sustained multi-core performance by the time retail firmware arrives, the 72,066 mark set by the 9950X3D becomes the target to chase.
AMD’s side of the story is less visible. The 9950X3D’s score is a known quantity, and no comparable leak for AMD’s next desktop generation appears in these reports. Our own look at AMD’s Ryzen AI Max Pro silicon against NVIDIA’s RTX Spark shows how aggressively the company is pushing in adjacent markets, so a quiet desktop roadmap should not be read as a lack of response.
How Benchmark Leaks Usually Age
Database leaks of unreleased processors have become a routine part of the hardware news cycle. They appear months before launch, they generate a burst of headlines and they rarely match final reviews. Early samples frequently underperform on multi-core tests, as this one appears to, and later improve once firmware matures. Occasionally the reverse happens, when a high-clocked sample leaks and retail parts ship with lower limits.
The smartest reading habit is to compare ratios rather than raw scores. A chip that beats its predecessor by 3% in one early leak and by 5% in a later one is trending in a useful direction. A single data point, like the 5,210 figure, is only a starting position. Outlets such as Notebookcheck framed this leak as promising, which fits that cautious stance.
Readers who remember earlier generations will recognize the pattern. A leak lifts expectations, a launch brings a mixed result, and the truth settles somewhere between the extremes. We cannot cite a specific past leak here without drifting beyond our fact sheet, but the cycle is familiar to anyone who covers the sector.
Market Impact for Buyers and Platforms
A Q1 2027 launch, if the claim holds, would put Nova Lake-S in stores within months of the new year. Nothing confirms that window, but it gives retailers and system builders a reason to watch inventory. A strong leak can also cool demand for current flagships, because buyers wait for the next generation.
Pricing is the biggest unknown. No source we reviewed reports a price, and chip costs depend on manufacturing and packaging capacity across the industry. Our coverage of TSMC weighing a 20% CoWoS price hike for NVIDIA chips shows how foundry and packaging economics are pushing costs up across the sector. Whether that pressure reaches desktop processors is something only Intel can answer.
Component budgets matter too. Storage is getting pricier, as we noted in the report on enterprise SSD prices rising 23% to 28% in Q4. A buyer eyeing a new CPU, board, memory and drive in early 2027 should budget for the whole platform, not just the processor.
What PC Builders Should Do Now
Do not cancel an upgrade you need today. A leaked sample with no price, no confirmed socket and no release commitment is not a reason to hold off on a purchase that solves a real problem. If your current machine runs well, waiting costs nothing.
If you are planning ahead, three habits pay off. First, avoid buying a motherboard for a chip whose socket is unconfirmed. Second, test your existing memory so you know your baseline, and our walkthrough on running Memtest86+ v9.00 on DDR5 shows how in 13 steps. Third, wait for independent reviews with retail hardware before spending money on any first-wave platform.
For more hardware news and buying guides, browse the hardware section of the site, where we track CPU, GPU and storage releases as they land.
Five Predictions for Nova Lake-S
These are our analysis, built from the reported numbers, and not confirmed facts. Treat each as a reasoned bet that later leaks or Intel’s own announcements could overturn.
- More entries will follow. Two 28-core results already exist, and PassMark’s database usually collects more samples as partners test boards. Expect additional listings with different core counts and clocks before any official reveal.
- The multi-thread score will rise. A 28-core part scoring below the 285K’s 67,274 looks like a firmware and power-limit story. We expect later samples to close part of the gap with the Ryzen 9 9950X3D’s 72,066, though we doubt they will fully erase it.
- Single-thread will stay near the current range. The tight spread between 5,210 and 5,191 suggests the design delivers consistent per-core speed. Retail numbers within a few percent of these are a fair expectation.
- Intel will not confirm specs before it is ready. We found no Intel statement on the entries, and silence usually holds until a formal launch event. Any claim of the Core Ultra 9 4970K BFC name will stay unofficial until then.
- Early reviews will decide the story. If a Q1 2027 launch holds, independent testing with final firmware will matter more than every leak combined. Buyers should plan around those results.
Frequently Asked Questions
What is the Nova Lake-S 28-core benchmark leak?
It is a set of PassMark PerformanceTest 11.1 entries, labeled “Intel 0000,” that outlets including TechPowerUp and VideoCardz tie to an early Nova Lake-S desktop processor with 28 cores. The best run scored 5,210 single-thread and 61,509 in overall CPU Mark.
Is the Core Ultra 9 4970K BFC an official product?
No. Reports offer that name as the likely identity of the sample, but Intel has not announced it. The name, socket, cache layout and retail specifications are all unconfirmed.
How does the leaked chip compare with the Core Ultra 9 285K?
In single-thread PassMark it scores 5,210 against the 285K’s 5,059, a lead of about 3.0%. In overall CPU Mark it posts 61,509 against 67,274, which is roughly 8.6% lower.
Does it beat the Ryzen 9 9950X3D?
Only in single-thread. The sample leads 5,210 to 4,805, about 8.4%. The Ryzen 9 9950X3D keeps a clear overall CPU Mark advantage at 72,066, about 14.6% above the best Nova Lake-S run.
When will Nova Lake-S launch and what will it cost?
A Q1 2027 launch was mentioned in reports, but Intel has not confirmed it. We found no confirmed price. Do not budget around either claim.
Can I trust leaked PassMark results?
Use them as a rough hint. Engineering samples run on pre-release firmware, and the final retail chip may score higher or lower. Database entries are also not official Intel disclosures, so wait for independent reviews.
Should I wait for Nova Lake-S before upgrading?
Only if your current system meets your needs and you can afford to wait. With no confirmed socket, price or date, upgrading today for a real need is the safer call. For background on current options, see the AM4 versus AM5 comparison.
