Updated Oct 20, 2026· 12 min read

Key takeaways

  • A very large gap between cores. If one core runs ten or more degrees hotter than the rest under a steady load, the cold plate probably does not touch evenly.
  • Temperature spikes straight to the limit. A processor that hits its limit within seconds of a light load usually has a dry patch, a missing film or a forgotten plastic cover.
  • Different temperatures after a reseat. If a second mount gives a very different result from the first with the same paste, the first mount was uneven.
  • Visible paste outside the spreader. A little is fine; a large amount means you used too much, and it should be wiped away before it reaches the socket contacts.

The best CPU thermal paste pattern for most processors in 2026 is a single pea-sized dot of about 3 to 4 mm in the center, and for the long rectangular heat spreaders on Intel LGA 1851 chips a thin line or a center dot with four small corner dots is a safe step up. Published comparisons show every common pattern landing within about two degrees of each other, so the amount of paste, even mounting pressure and a flat cold plate matter more than the shape you draw.

Our top picks

As an Amazon Associate we earn from qualifying purchases at no extra cost to you.

Quick Comparison of Seven Paste Patterns

Pattern Best for Typical amount Measured result (Puget Systems data) Main risk
Center dot (pea) AM4, AM5, most square heat spreaders 3 to 4 mm blob 54.75 C Thin edges on very large spreaders
X shape High-wattage chips, extra corner coverage Two thin crossing lines 54.25 C Easy to overapply
Thin line Rectangular LGA 1851 and LGA 1700 spreaders One line along the long axis Not published as a separate result; thick line measured 56.00 C Too thick a line leaves a poor result
Five dots (center plus corners) Large rectangular spreaders 3 to 4 mm center, 2 mm corners Comparable to the dot and X Small dots can trap air if misplaced
Spread with a card Perfectionists, large Threadripper-class spreaders Thin, even film 54.50 C Air bubbles, contamination from the spreader tool
Four-quadrant dots Very large chips Four small dots Similar to other patterns Under-coverage of the middle
Thick line Not recommended Heavy line 56.00 C Poor spread and paste squeeze-out

The measured values come from a Puget Systems comparison, repeated by Sunbeam; a second source gives a slightly different dot figure of 55.1 C. All patterns fell within roughly 1.75 C of the best result.

Why the Pattern Matters Less Than You Think

When you tighten a cooler onto a processor, the clamping force spreads the paste outward from wherever you placed it. The job of the paste is to fill the microscopic gaps between the metal of the heat spreader and the cold plate, not to form a thick layer; a thick layer is actually worse than a thin one because paste conducts heat far worse than copper. A good application leaves a film only a few hundredths of a millimeter thick, with the surplus squeezed out at the edges.

GamersNexus ran a comparison of application methods on a delidded processor drawing around 256 W and found a spread of less than one degree from best to worst, as reported secondhand by Ziitek. Guidespot’s figures for a Ryzen 7 9800X3D showed 65.8 degrees for a dot and 64.5 degrees for an X, a gap of about a degree. None of these are large, which is why the quantity of paste and the quality of the mount are the details worth getting right.

Best Overall: The Center Dot

A round dot of 3 to 4 mm in the exact center of the heat spreader is the pattern most guides, including Alibaba’s buying guide and Tom’s Hardware’s how-to, recommend for typical processors. When the cooler’s cold plate presses down, the dot spreads into a circle that covers the hot area directly above the die. On AM5 chips, the heat spreader has two cutouts near each edge, so the usable area is relatively small and a center dot fits it well.

Who it suits: first-time builders and nearly every AM4 and AM5 Ryzen processor.

Pros: simple, repeatable, almost no risk of air pockets, and little cleanup.

Cons: on a large rectangular spreader, the ends can receive a thinner film.

Best for High-Wattage Chips: The X Pattern

Two thin crossing lines draw an X from corner to corner. In Puget Systems’ data the X gave the lowest reading, 54.25 C, narrowly ahead of a spread and a dot. The shape sends paste toward all four corners, which helps chips with a large heat spreader or several hot spots. Guidespot’s 9800X3D figures also showed the X ahead of a dot by about a degree.

Who it suits: owners of 120 W or higher chips and people who want slightly more corner coverage.

Pros: excellent edge and corner coverage and the best measured number in the Puget comparison.

Cons: it is easy to use too much paste, which then squeezes out onto the socket area.

Best for Rectangular Spreaders: The Line

Intel’s LGA 1700 and LGA 1851 processors have a rectangular spreader, and the die sits in the middle along the long axis. A thin line along that axis puts paste where the heat is. Several sources favor a line or an X for these chips, while Tom’s Hardware suggests a different layout, a single center dot of 3 to 4 mm plus four 2 mm dots set a couple of millimeters in from each corner. Sunbeam’s own guides disagree on whether a plain dot suffices, noting that the corners of the spreader usually run cooler than the center anyway.

Who it suits: Core Ultra 200S owners and Intel 12th to 14th generation users.

Pros: covers the long axis well, minimal squeeze-out if kept thin.

Cons: a thick line measured worst in the Puget data, at 56.00 C, so thinness matters.

Best for Large Chips: The Five-Dot Layout

This is the Tom’s Hardware pattern for LGA 1851: one 3 to 4 mm dot in the center and four 2 mm dots near the corners. It is a conservative way to guarantee coverage over a long spreader without drawing lines. On AM5 it is unnecessary, and it risks trapping air if the corner dots are too close to the edge.

Who it suits: Intel builders who want a standard method with predictable results.

Pros: reliable coverage, easy to repeat.

Cons: more steps than a single dot and a small chance of misplaced dots.

Best for Precision: Spreading with a Card

Spreading the paste into a thin, even film with a plastic card or the tool included with the tube produced a 54.50 C result in the Puget comparison, between the X and the dot. The method gives complete coverage but carries more risk: air bubbles, contamination from a dirty card and uneven thickness are all possible. It makes sense when you work with a large spreader, such as a workstation chip, or when you simply like to see a perfect layer.

Who it suits: experienced builders and large-spreader platforms.

Pros: total coverage and no dependence on the squeeze pattern.

Cons: easy to get wrong, slower, and Sunbeam notes that a slightly larger dot or a pre-coat tends to improve edge coverage without the full spread.

Patterns to Avoid: Thick Lines and Excess Paste

The one clearly poor result in the Puget data was the thick line at 56.00 C, about 1.75 C worse than the best result. The worry with too much paste is rarely temperature alone; excess paste squeezes out and can contaminate the socket area, and on some boards it can reach exposed contacts. Non-conductive pastes do not short anything, but they are messy to clean. Liquid metal is a different story and is covered below.

How Much Paste: A Worked Calculation

A typical heat spreader measures about 40 mm by 40 mm on AM5, an area of 1,600 square millimeters. A well-mounted film is roughly 0.03 mm thick, giving a volume of about 48 cubic millimeters, or 0.048 milliliters. A 3.5 mm sphere has a volume of about 22 cubic millimeters, which spreads to a film of about 0.014 mm over the full area, but in practice the contact area is smaller than the full spreader, and the real film is therefore thicker. A 4 mm sphere holds about 33 cubic millimeters. So a 3 to 4 mm dot is in the right range, and it is why a larger blob only produces squeeze-out rather than a thicker working layer.

A standard 2 g tube holds about 0.5 to 0.7 milliliters, depending on density, enough for roughly 15 to 20 applications of a 4 mm dot. That means a single small tube covers many builds, and it is a reason not to apply a heavy hand.

Step by Step: Applying Paste for the Best Result

  1. Remove the old paste. Use a lint-free cloth or a coffee filter with isopropyl alcohol of 90% or higher on both the heat spreader and the cold plate, because old paste carries dust that creates gaps.
  2. Let it dry. Alcohol evaporates quickly; wait about a minute before applying new paste.
  3. Choose the pattern from the table above. A center dot for AM4 and AM5, and a line or five-dot layout for LGA 1700 and LGA 1851.
  4. Apply a small amount. Start with a 3 mm dot. You can add more on a reseat, but cleanup after too much is slow.
  5. Lower the cooler straight down. Do not slide it sideways, which drags air into the film.
  6. Tighten evenly. Tighten opposite screws a little at a time in a cross pattern, then fully tighten, so pressure is even.
  7. Check temperatures. Compare an idle reading and a gaming load reading with your earlier log. Some pastes need a short settling period, so recheck after a day or two.

Pattern Choice by Socket and Situation

Socket or situation Spreader shape Recommended pattern Notes
AM4 (Ryzen 5000 and earlier) Square Center dot Same cooler mount as AM5
AM5 (Ryzen 7000 and 9000) Square with edge cutouts Center dot; X for 120 W chips Slightly larger dot if you want edge coverage
LGA 1700 (12th to 14th gen) Rectangular Line along the long axis or X Contact frame may bend the spreader
LGA 1851 (Core Ultra 200S and Plus) Rectangular Center dot plus four corner dots, or a thin line Tom’s Hardware pattern
Reusing an old cooler Any Fresh paste, same pattern Always clean both surfaces
Liquid metal Any Tiny amount spread on both surfaces Not a paste pattern; see safety notes

Paste Viscosity and Why It Changes the Advice

Pattern advice assumes a typical medium-viscosity paste such as those from Arctic, Noctua or Thermal Grizzly. A very thick paste spreads less under pressure, so it benefits from a slightly larger dot or an X, while a thin, runny paste spreads easily and punishes overapplication with greater squeeze-out. Check the manufacturer’s own instructions in the box; they usually state the intended amount, and some include a spreader tool tuned to that paste. Where those instructions and general advice differ, follow the maker, because they know the compound’s flow behavior.

Some compounds also need a short settling period before they reach their best performance, as the paste thins slightly under heat and the film reaches its final thickness. If your first reading looks a degree or two high, run a few heat cycles through normal gaming before deciding the application was poor.

Mounting Pressure Is the Hidden Variable

The same dot can give very different results depending on how the cooler is fastened. Most modern coolers use spring-loaded screws that stop at a preset pressure, which is the easiest way to get even contact. Backplate-and-standoff designs depend on you tightening each screw in turn. Always make half-turns across the diagonal, then finish each screw to its stop. Never tighten one corner fully before touching the others, because that tilts the cold plate and leaves a wedge-shaped film.

The cold plate itself matters as well. Reviewers who measure cooler base flatness find that some bases are slightly convex by design, which concentrates pressure at the center and is one reason a center dot works so well. A rough or visibly machined base can leave channels that the paste must fill, so a marginally larger dose helps. If a base looks scratched or uneven, do not lap it yourself unless you are experienced, because removing the plating can create problems of its own.

Signs Your Application Went Wrong

  • A very large gap between cores. If one core runs ten or more degrees hotter than the rest under a steady load, the cold plate probably does not touch evenly.
  • Temperature spikes straight to the limit. A processor that hits its limit within seconds of a light load usually has a dry patch, a missing film or a forgotten plastic cover.
  • Different temperatures after a reseat. If a second mount gives a very different result from the first with the same paste, the first mount was uneven.
  • Visible paste outside the spreader. A little is fine; a large amount means you used too much, and it should be wiped away before it reaches the socket contacts.

Check these with a monitoring tool such as HWiNFO by logging an idle period followed by a steady gaming load. Compare the result with a build from a similar processor and cooler as a reference, bearing in mind room temperature changes every number.

Liquid Metal and Other Special Cases

Liquid metal compounds are a separate case, because they are electrically conductive and cannot be applied as a pattern on top of a heat spreader the way a paste can. They are spread as a very thin coat on both the heat spreader and the cold plate, and they must never touch aluminum, which they corrode. Thermal Grizzly recommends nickel-plated copper surfaces for best long-term stability. The surrounding area should be protected, and the technique is for experienced builders only. If you are not sure, a high-quality ceramic or carbon paste is the safer choice.

Phase change pads are another special case: they come cut to size, so there is no pattern to choose at all. They are solid at room temperature and soften once the processor warms them, then fill the gaps on their own.

Common Mistakes

  • Too much paste. A heap of paste does not improve conduction and it makes cleanup harder.
  • Too little paste. Dry patches over the die cause hot spots. A dot that spreads to cover the middle of the spreader is the minimum.
  • Leaving the plastic film on the cooler. This is the classic first-build error; check the cold plate before mounting.
  • Uneven tightening. A tilted cooler leaves a wedge-shaped film, hot on one side.
  • Reusing old paste. After a cooler is lifted, the film is broken, so clean and reapply.
  • Skipping the pattern for LGA sockets. A plain pea works, but on a rectangular spreader, a line or five dots is the safer choice.

Frequently Asked Questions

What is the best thermal paste pattern for a Ryzen CPU?

A single center dot of 3 to 4 mm works well for AM4 and AM5 processors. For chips drawing 120 W or more, an X pattern adds corner coverage and measured slightly cooler in the comparisons cited above. Either choice is within a degree or two of the other.

Does the dot, X or line pattern change temperatures much?

Not by much. The Puget Systems comparison placed every common pattern within about 1.75 degrees Celsius of the best, and other analyses report even smaller gaps. Mounting pressure and the quantity of paste have a larger effect.

How much thermal paste should I use?

Use a pea-sized dot of about 3 to 4 mm for a square spreader. Larger amounts only squeeze out. If you see paste oozing well beyond the spreader edge after mounting, you used more than needed, though it is harmless with a non-conductive paste.

Should I spread the paste by hand?

You do not need to. Spreading gives complete coverage but risks air bubbles and contamination, and the measured benefit is negligible. If you prefer it, use a clean plastic card and a thin, even film.

How often should I replace thermal paste?

Every three to five years is a common interval for standard compounds, sooner if temperatures climb at the same workload. Replace it whenever you remove the cooler, since lifting the cooler breaks the film.

Can I use the same pattern on Intel and AMD?

You can, and a center dot will work on both. Intel’s rectangular spreaders on LGA 1700 and LGA 1851 benefit from a line or five-dot layout, which gives the long ends of the spreader a bit more coverage.

A
affadmin
We compare specs, materials and verified owner reviews before a product earns a spot. Rankings are never paid.
Affiliate disclosure. As an Amazon Associate we earn from qualifying purchases at no extra cost to you. Prices accurate as of the date shown.
Best CPU Thermal Paste Pattern in 2026: Dot,…Check price on Amazon