Still Hot After New Thermal Paste? How Case Airflow, Intake/Exhaust & Pressure Affect GPU Temperatures

2026-10-19 ｜ GPU Repair ｜ 12 min read ｜ HK GPU Fix Editorial

Key Takeaways

New thermal paste only fixes heat transfer between the GPU die and the cooler base. If the cooler cannot shed heat into case air, or case air cannot leave the case, temperatures stay high. This guide explains how airflow, intake and exhaust, and pressure affect GPU temperatures, and how to test changes one at a time.

Why New Thermal Paste Does Not Always Lower GPU Temperatures

It is a frustrating result: you replace the thermal paste, reassemble the card carefully, and the temperature under load barely moves. Before assuming the paste job failed, it helps to picture the full heat path.

Heat from a graphics card travels through several links in a chain:

1. From the GPU die into the cooler base, through the thermal paste or other thermal interface material.
2. From the cooler base into the fins of the heatsink.
3. From the heatsink fins into the air passing over them.
4. From the air inside the case to the air outside the case, carried out by the case fans.

New thermal paste only addresses the first link. It does nothing for the third and fourth links. If the heatsink fins cannot pass heat into the air, or if the warmed air inside the case cannot leave the case, the GPU will stay hot no matter how fresh the paste is. This is a statement about how the heat path is divided, not a claim that replacing paste is useless. Paste matters when the interface is the weak link. Airflow matters when the air side is the weak link. The diagnostic question is therefore not "did I apply the paste correctly?" alone, but "which link in this chain is actually limiting cooling right now?"

The rest of this guide works through that question: how a graphics card cooler interacts with case air, how intake and exhaust fans shape that air, how positive and negative pressure change dust and heat behaviour, which layout mistakes are common, how to test a change properly, and where the boundary lies between an airflow problem and something else entirely.

How a Graphics Card Cooler Actually Uses Case Air

Most consumer graphics cards today use an open-air cooler design. According to How-To Geek and AORUS, the card's fans draw air from inside the case, push it across the heatsink fins, and then release the warmed air back into the case from the top, bottom, or sides of the card. The case fans must then remove that warmed air from the case. An open-air card is therefore highly dependent on case airflow. Its cooler can only work with the air the case gives it.

A blower-style cooler works differently. According to the same sources, a blower fan draws in case air and pushes it through a sealed shroud so that the warmed air exits directly out of the rear I/O bracket, rather than being released inside the case. A blower card is comparatively independent of case ventilation, but it is generally louder and uses fewer fans. Blower designs are now uncommon on new consumer cards, so most readers will be dealing with an open-air card whose cooling is tied to the case.

This distinction explains a central mechanism. As a matter of general cooling principle, the rate at which a heatsink sheds heat depends broadly on the temperature difference between the fin surface and the air flowing over the fins. The hotter the air arriving at the fins, the less heat the same heatsink can remove. If the case air has already been warmed by the GPU and CPU, if the room itself is warm, or if warmed air is not being exhausted, the GPU runs hotter. This is a qualitative mechanism. It does not support any precise conversion such as "a one-degree rise in intake air means a one-degree rise in GPU temperature," and no such conversion is offered in this guide.

A related mechanism is recirculation. Warmed air released by an open-air card, if it is not carried out of the case by the exhaust fans, can be drawn back in by the GPU fans. The card then cools itself with air that has already been heated, which can form a self-reinforcing loop. Again, this is a qualitative description of a mechanism. There is no reliable general figure for how much recirculation raises temperature, and none is given here.

Intake and Exhaust: The Basics Done Right

Noctua's airflow guide describes the basic job of case cooling as continuously supplying cool air to every component while removing warm air at roughly the same rate. Intake fans pull fresh air from outside the case inward. Exhaust fans push warm air from inside the case outward. That sounds simple, but several details decide whether the arrangement actually delivers cool air to the graphics card.

One direction per face

Noctua suggests that beginners keep each face of the case to a single direction — all intake or all exhaust on that face — rather than mixing intake and exhaust on the same face. As an example of a basic arrangement for a case with six fan positions (three front, two top, one rear), Noctua describes two front intake fans plus one rear exhaust fan. The point is not that this exact count suits every case. The point is the discipline: air should have a clear path in, across the hot components, and out.

Fan speed, size, and count

Noctua's guide also notes the basic trade-offs of fans themselves. Higher fan speed moves more air but produces more noise. At the same airflow, a larger fan can run at a lower speed and be quieter. More fans can each run at a lower speed while delivering the same total airflow. These trade-offs matter when tuning a case: adding airflow does not have to mean accepting much more noise, and raising every fan to full speed is not the only way to move more air.

Give the front panel room to breathe

Noctua advises that a case should have a relatively open mesh face, especially on faces with fan mounts. An intake can use a dust filter, but the filter will reduce airflow to some degree. Noctua describes glass or closed front panels as severely restricting intake and as a major cause of high case temperatures. The lesson is structural: if the front panel barely lets air in, the intake fans behind it cannot do their job, however fast they spin.

Noctua's checklist for intake includes an open front panel, such as a coarse mesh, with no filter or only a low-resistance filter. Filter density and the distance between the filter and the fan matter. A filter that is too dense will clog faster, and once clogged it resists intake air even more. Noctua suggests a simple feel test: blow through the filter and compare the airflow you feel with and without the filter in place. That gives a rough sense of filter resistance, not a measured airflow figure.

Noctua also suggests that a standard mid-tower case should have at least five 120 mm fan positions. That is a statement about fan positions, not a requirement to fill all five with fans. A fan position is itself an opening through which air can pass freely.

Keep cables out of the wind

Cable management matters for airflow. Noctua notes that tidying cables reduces airflow obstruction and reduces audible airflow turbulence. In practice this means routing cables behind the motherboard tray or tying them down, rather than letting them cross the main path from the front intake to the GPU and CPU.

Large cards in smaller cases need more attention

Noctua notes that a graphics card longer than 300 mm, or thicker than three slots, can be harder to cool in some cases, particularly smaller ones. In such situations, a bottom intake fan that supplies fresh air directly to the graphics card, a mesh side panel, or a side fan can be a great help to the GPU. The underlying idea is the same as throughout this guide: the card needs a supply of air that has not already been heated, delivered close to where its fans can use it.

What Noctua's example fan combinations do and do not prove

Noctua's second airflow guide describes fan combinations tested in one specific case, the Fractal Design North. In that case, a five-fan combination of three front intake fans, one top intake fan fitted with an intake spacer, and one rear exhaust fan produced the lowest average temperatures for the CPU, GPU, and case in that test. The lowest front fan continuously blew air toward the graphics card, which benefited the GPU. A six-fan combination in the same case — three front intake, one rear exhaust, one front-top intake with a spacer, and one rear-top exhaust — performed best in that test.

These are results for one specific case. They do not establish a universal rule such as "the front-top fan must always be intake." Every case has a different panel layout, filter arrangement, and internal volume, so a combination that works well in one case may not be the best combination in another. Treat these examples as illustrations of how to think about airflow paths, not as a prescription for all cases.

The same guide gives a concrete example of an intake-exhaust short circuit. The front-most top exhaust fan can pull in fresh air from the front-most top intake fan and expel it before that air ever reaches the CPU cooler, leaving the CPU cooler short of fresh air. This is the pattern to watch for: fresh air that enters the case and leaves again without passing over a hot component has done no useful cooling.

Positive, Negative, and Balanced Pressure

Pressure is one of the most misunderstood ideas in case cooling. It is worth defining the three cases carefully, because none of them is universally best.

These definitions and trade-offs come from Noctua's airflow guide, which also notes that all three configurations have their own strengths and weaknesses and should be chosen according to the situation. That caution matters: claims such as "negative pressure is always cooler" or "positive pressure is always better" are not supported. Positive pressure has a heat-accumulation drawback, and negative pressure has a dust drawback.

How pressure is actually controlled

Noctua notes that, without changing the case structure, the main way to control case pressure is the ratio of intake to exhaust fans. In general, a difference of one or two fans between intake and exhaust is already enough to affect the pressure inside the case. Fan speed also changes pressure: Noctua gives the example that with two intake and two exhaust fans, raising intake speed can create positive pressure, while raising exhaust speed can create negative pressure. Counting fans alone therefore does not determine pressure. Fan speed, fan specification, filters, and front-panel resistance all change the actual airflow.

Noctua considered a positive-pressure configuration advantageous in its example case. That is a finding about its example case. It cannot be rewritten as "positive pressure is always best" or "positive pressure always makes the GPU cooler," because positive pressure carries the warm-air accumulation drawback described above.

How an ordinary user can judge pressure, roughly

Most users have no instrument for measuring the actual air pressure inside a case. Three practical checks can give a qualitative indication of the tendency, and all three must be described as tendencies rather than measurements:

• Compare the rated airflow of the intake and exhaust fans along with their actual speeds. This is only an estimate, because filters, the front panel, cables, and fan-position resistance all change the real airflow.
• The tissue-paper test: hold a small piece of thin paper near an un-fanned opening or gap, switch the machine on, and observe whether the paper is pushed outward, which suggests positive pressure, or drawn inward, which suggests negative pressure. This is a qualitative indication, not a measurement.
• Observe the dust pattern. Dust collecting mainly at filtered intake positions suggests positive pressure with filtered intake. Dust entering through unfiltered gaps and various openings suggests negative pressure.

None of these methods measures pressure. They indicate a tendency. A common recommendation is slight positive pressure — intake slightly greater than exhaust — as a way of balancing dust control and cooling. But "slight" has no agreed numerical definition, and no figure such as "intake must exceed exhaust by a set percentage" should be invented. Noctua's "one to two fans" remark describes its example case, not a universal formula.

Common Layout Mistakes

The following mistakes are the ones most often found when a card stays hot after a paste change. Each is described with its mechanism and its fix, without any promise about how much temperature will change.

A fan installed backwards

According to PCWorld, most fan frames carry two arrows on the side: one indicating the direction of blade rotation, and one indicating the direction of airflow. The side the airflow arrow points toward is the exhaust side. When there is no arrow, the convex side of the blades — the rounded, outward-curving side — is the intake side, and the concave, bowl-like side is the exhaust side. The side with the support struts or cross-brace and the central label sticker is usually the exhaust side. Readers should check the arrows with the machine switched off rather than relying on any single feature such as the sticker, because label placement varies between manufacturers and the arrow is the most reliable indicator. A single reversed fan can quietly undo an otherwise sensible layout, for example by turning a planned intake into an extra exhaust that starves the GPU of fresh air.

Intake-exhaust short circuit

A short circuit happens when fresh air enters and leaves the case without passing over the main hot components. Noctua's example of a front-top exhaust fan removing fresh air before it reaches the CPU cooler is one instance. Mixing intake and exhaust on the same face is another. A further variant is a path in which front intake air flows straight to a top or rear exhaust without passing the GPU or CPU at all. The fix is to think in paths rather than in fan counts: every intake should have a reason to pass over a hot component before it reaches an exhaust.

A sealed front panel

A glass or sealed front panel severely restricts intake. In one test-platform comparison by Gamers Nexus, a sealed front panel substantially restricted intake, while removing an already-mesh front panel gave only a limited improvement; that result belongs to that specific test and cannot be generalised to all sealed panels. The improvement direction is to fit a mesh front panel if the case offers one, or to accept that case's intake limitation, rather than assuming that adding more fans will necessarily solve the problem. Before fitting a mesh panel, confirm that the case has an official mesh panel option. Do not drill holes or modify the panel yourself, since modification may affect warranty and structure.

A clogged filter or front panel

A filter that is too dense, or a filter choked with dust, increases intake resistance. Filters should be checked and cleaned regularly. No sourced figure exists for how often, so no cleaning interval is given here. The check itself is simple: look at the filter, feel the airflow with and without it as Noctua suggests, and clean it when it is visibly loaded with dust.

Cables blocking the path

Loose cables across the main airflow path obstruct air and add turbulence. Route cables behind the motherboard tray or tie them down, keeping the path from the front intake to the GPU and CPU clear.

A graphics card pressed against the case floor or PSU shroud

In a small case, a large card can end up with its fans almost touching the PSU shroud or the case floor, so the fan inlets are obstructed and the fans cannot draw enough air. No authoritative source was found that sets a general minimum clearance figure for this gap, so no millimetre figure is given here. The description stays qualitative: if the GPU fans are nearly touching the shroud or the floor, the intake is obstructed. A bottom intake fan or a mesh side panel can help in such a case, as noted above.

Vertical mounting

Vertical mounting changes where the card's fans face and where its air comes from, so its effect depends on the case. According to Corsair's test in the AIR 5400 case, vertically mounting the GPU made GPU temperatures 2°C to 4°C higher than a conventional horizontal mount in that test, with results varying by GPU cooler design, while VRM, RAM, and M.2 temperatures were slightly lower, because the space behind the GPU opened an additional upward airflow path. Corsair also noted that results vary with GPU cooler design. Both directions must be reported together: it is not accurate to say only that vertical mounting makes the GPU hotter. After vertical mounting, the GPU fans face the side panel instead of facing the bottom intake. If that side panel is glass and sits very close to the fans, intake is restricted as a matter of general principle. No authoritative source was found that sets a general minimum clearance between the fans and the glass, so no millimetre figure is given.

A riser cable itself should not be blamed for temperature. No authoritative source was found showing that a riser cable directly makes a GPU hotter. The main risks a riser brings are signal integrity and compatibility — PCIe generation, shielding, and length — along with the fact that the card ends up in a different position facing a different direction. Anyone fitting a riser should follow the case and motherboard documentation to confirm PCIe generation compatibility, since a poor-quality or mismatched riser can cause signal problems. That is a general installation caution, not a cooling conclusion, and no temperature figure is attached to riser cables.

PSU fan direction

According to Huntkey, the principle for PSU fan direction is that the PSU fan should face the nearest ventilation opening. In a case with a bottom vent and filter, the PSU fan faces downward, drawing cool air directly from outside the case, with exhaust leaving through the rear, so PSU heat does not enter the case. In a case without a bottom vent, the fan faces upward, drawing warmer air from inside the case, which is still better than drawing no air at all. A side effect of the upward orientation is that the PSU then competes with the GPU for the air inside the case. If the machine sits on carpet or another surface that blocks the bottom intake, bottom intake will be obstructed. Placement matters as much as orientation.

Where the case sits

If the front or side intake is pressed against a wall, if the case is pushed into a closed cabinet, or if objects are piled in front of it, intake is obstructed. If the rear or top exhaust is pressed against a wall, warm air lingers and can be drawn back in. These are direct applications of the airflow-path idea. Checking that no intake or exhaust opening is blocked is a legitimate inspection step, and it needs no measurements.

Dust in the heatsink fins

Dust packed between the heatsink fins obstructs air passing through the fins and reduces the effective cooling area. Strictly speaking this is a problem with the cooler itself rather than with case airflow, but readers easily confuse the two. It is therefore covered in the boundary section below rather than treated as an airflow problem.

How to Test: Before-and-After Comparisons with GPU-Z and HWiNFO

Guessing is the enemy of airflow tuning. The way to know whether a change helped is a controlled before-and-after comparison, using the same load, the same room conditions, and the same fan settings.

Which readings to watch

According to BreakFixNow, HWiNFO in Sensors-only mode can show GPU temperature, GPU hotspot temperature, GPU memory temperature, GPU fan speed in RPM, power draw, and clocks at the same time. GPU-Z's Sensors tab can show temperature, clocks, fan speed, and memory usage. Windows Task Manager shows only the GPU core temperature, with no hotspot, memory temperature, or fan speed, which is not enough for diagnosis.

GPU-Z version 2.71.0 added hotspot temperature and hottest-memory-chip temperature readings for NVIDIA RTX 50 series cards, according to OC3D, along with additional clock readings. Different models, generations, and manufacturers expose different sensors, so readers should not assume that every card provides hotspot and memory temperature readings. Check what the installed card actually reports.

For AMD, one manufacturer statement needs careful scoping. AMD said that RX 5700 series cards keep raising clocks until one sensor reaches 110 °C junction or hotspot temperature, and that reaching 110 °C junction temperature in ordinary gaming use is expected and within specification for that series, according to AMD's statement as reported by bit-tech. This applies to the RX 5700 series only. It cannot be generalised to "all AMD cards are fine at 110 °C," and it cannot be applied to NVIDIA cards. Readers should check the official specification for their own model rather than applying another model's figure.

Fans that stop at idle are often normal

According to ASUS support documentation, its 0dB technology stops the GPU fans when the GPU temperature meets the relevant condition, and restarts them automatically as temperature rises. A fan that does not spin at idle is therefore not by itself evidence of a fault. Judge fan behaviour under load: check whether the fans spin and whether speed rises with temperature. ASUS also advises that if the fans still do not spin once the fan-start condition has been reached, the driver should be updated, the machine should be shut down and disconnected, and dust blocking the fans should be cleaned. If a fan is severely stuck, the card should not continue to be used with a faulty fan, and official support should be contacted.

Fix the variables before comparing

A before-and-after comparison is meaningless unless the variables are fixed. Each comparison should hold the following constant:

• The same load: the same game or the same benchmark, the same scene or settings, and the same duration.
• The same room temperature, recorded each time. EKWB's explanation of delta T — measured temperature minus ambient temperature — shows why this matters: when room temperature changes, measured temperature changes with it. Comparing two GPU temperatures taken on different days at different room temperatures proves little, and delta T is the concept that explains why.
• The same fan settings: do not compare one run on an automatic fan curve with another run on manual full speed.
• The same set of readings: GPU core temperature, hotspot temperature, memory temperature where available, GPU fan RPM, power draw, and clocks. Watching a single "GPU temperature" figure is not enough.

The side-panel test: a coarse screen, not a diagnosis

Opening the side panel and repeating the same load is a useful coarse screen. If temperatures clearly fall with the panel open, that tends to point to restricted case intake or exhaust, because opening the panel bypasses the front panel, filter, and airflow-path restrictions. If temperatures do not clearly change with the panel open, that tends to point to a non-airflow cause: the cooler itself, the paste or pad installation, a faulty fan, room temperature, or a power limit, as set out in the boundary section below.

The side-panel test has a hard limit that must be stated plainly: it cannot tell you which fan is reversed or which position is blocked. It is a coarse screen only, not a confirmation of a specific fault. Opening the side panel is also not a solution. It is a diagnostic comparison, and the machine should not be left running with the panel open long-term, since that lets dust enter directly and changes the airflow path.

Practical points for the side-panel comparison: run the same load until temperatures stabilise, before and after opening the panel — a short run is not enough. Change only one variable at a time: only open the panel, only change a fan direction, or only adjust a fan curve. Allow temperatures to settle back down between changes. Changing two things at once makes it impossible to tell which change helped.

Reading fan speed

Under the same load, high fan speed together with high temperature can mean the cooling system is chasing temperature — for example because the intake air is hot or the fins are dusty. Low fan speed together with high temperature calls for a check of whether the fan curve is too conservative, whether a 0dB mode is holding the fans back, or whether a fan is faulty. These readings are clues that call for further checks, not diagnoses in themselves.

Reading power and clocks

Under the same load, if power draw has already reached the card's power limit while temperature has not reached that model's throttling temperature, and clocks are limited, the limit may be a power limit rather than a temperature limit. If temperature reaches the throttling temperature and clocks fall at the same time, that tends to indicate a temperature limit. The word "tends" matters. Readers must check their own model's power limit and temperature specification, and no general throttling temperature figure is given here, because throttling temperatures vary by model.

Improvement Steps, from Free to Costly

Work in order: free checks and tidying first, then settings, then parts. After each step, run one before-and-after comparison using the method above, and confirm that the step helped before moving on.

1. Check that no intake or exhaust opening is blocked, and check where the case sits. Keep the front and side intakes away from walls, out of closed cabinets, and clear of objects piled in front. Leave room behind and above the case for exhaust air.
2. Check and clean the filters, fan blades, and the surface dust on the heatsink fins, with the machine switched off and disconnected. A clogged filter increases intake resistance. No sourced cleaning interval exists, so no schedule is prescribed.
3. Tidy the cables. Route them behind the motherboard tray or tie them down, keeping the main path from the front intake to the GPU and CPU clear.
4. Check every case fan direction against the arrows on the fan frames, and correct any reversed fan. Then run one before-and-after comparison.
5. Check the fan curves and settings. Fan speed should rise with the temperature of the key components. If exhaust speed is persistently far above intake speed, the case may tend toward negative pressure; if intake is persistently far above exhaust, it may tend toward positive pressure. After adjusting, run another before-and-after comparison. Do not change several settings at once.
6. Adjust the fan configuration for the case. A common starting point is front intake plus rear exhaust — Noctua's basic three-fan example is two front intake plus one rear exhaust, and its five-fan example adds further intake and exhaust positions as described earlier. Call it a starting point, not a rule that makes every case coolest.
7. Consider adding a fan if the case has an empty fan position. Adding a fan can let each fan run at a lower speed for the same total airflow, which helps noise. Which position to fill depends on the case manual's fan positions. A bottom intake helps a large GPU, as noted above.
8. Fit a mesh front panel if the case has that option. A sealed or glass front panel is a major intake restriction. Confirm that the case has an official mesh panel option before proceeding, and do not drill or modify panels yourself.
9. Improve the ventilation of the room around the machine. Lower room temperature makes it easier for a heatsink to maintain its temperature difference. Avoid placing the machine in a closed cabinet, beside a heater outlet, or in direct sunlight. No figure is given for how much GPU temperature changes per degree of room temperature, because no sourced conversion exists.
10. Replace the case as a last resort. If the case has too few fan positions, a sealed front panel with no mesh option, and too little room for a large card, a case with enough fan positions and a mesh front panel is the structural answer. Replacing the case only provides better airflow conditions; the fans must still be configured correctly using the steps above.

No step in this list carries a promised temperature improvement. No figure such as "this step typically drops temperatures by several degrees" is given, because no general figure is sourced. Figures from any single test belong to that test's specific case, card, and conditions, and cannot be reused as general improvement claims.

When It Is Not an Airflow Problem

The side-panel test points in one of two directions. If opening the panel clearly helps, airflow is the likely limit. If it does not, the cause is more likely elsewhere. The table below sets out the main alternatives, with the clues that tend to point toward each. Every entry is a tendency, not a confirmation.

The key boundary sentence is worth stating exactly: if temperatures clearly fall with the side panel open, the case airflow tends to be the limit; if temperatures do not clearly change with the panel open, the cause tends to lie in the cooler itself, the installation, a fan, room temperature, or power, and each needs to be checked in turn.

FAQ

If new thermal paste did not help, does that mean the paste job was bad?

Not necessarily. Paste only fixes the interface between the die and the cooler base. If the limit is on the air side — fins that cannot shed heat, or case air that cannot leave — a good paste job will still leave temperatures high. Run the side-panel comparison and check the airflow steps before reopening the card.

How can I tell which way a case fan is blowing?

Switch the machine off and look at the arrows on the fan frame. The airflow arrow points toward the exhaust side. If there is no arrow, the convex blade side is generally the intake side, the concave side is the exhaust side, and the side with the support struts and central label is usually the exhaust side. The arrow is the most reliable indicator.

Is positive or negative pressure better for GPU temperatures?

Neither is universally better. Positive pressure helps keep dust out of gaps but can let warm air accumulate. Negative pressure pushes warm air out but draws dust in through gaps. Balanced pressure is a compromise. Noctua describes all three as having strengths and weaknesses to be chosen by situation, and considered positive pressure advantageous in its example case only.

Should I just leave the side panel off?

No. Opening the side panel is a diagnostic comparison, not a fix. Running long-term with the panel off lets dust enter directly and changes the airflow path. If opening the panel clearly lowers temperatures, use that result as evidence of an airflow restriction, then fix the restriction — blocked openings, a clogged filter, a reversed fan, or a sealed front panel — and put the panel back on.

My GPU fans do not spin at idle. Are they broken?

Usually not. Many cards use a 0dB feature that stops the fans at low temperature and restarts them as temperature rises, as described in ASUS support documentation. Judge fan behaviour under load. If the fans still do not spin once the fan-start condition has been reached, update the driver, shut down and disconnect the machine, clean dust blocking the fans, and if a fan is severely stuck, stop using the card with the faulty fan and contact official support.

The core temperature looks fine but the hotspot is high. What does that mean?

It tends to suggest a contact or interface problem in the heat path from the die through the cooler, rather than a case-airflow problem. Do not apply a general degree threshold for the gap, because none is sourced. Compare the same load before and after any paste or pad work, check the installation against the documentation for the specific card model, and remember that sensor availability varies by model and generation.

Why should I not compare temperatures taken on different days?

Because room temperature changes measured temperature. EKWB's delta T concept — measured temperature minus ambient temperature — explains why. Two GPU temperatures taken at different room temperatures are not directly comparable. Record the room temperature each time, keep the load and fan settings fixed, and compare like with like.

Can I trust a single review's temperature improvement figure?

Only for that review's conditions. A figure from one case, one card, and one test setup describes that test. It cannot be reused as a general claim about what adding a fan, changing a panel, or moving a card will do in a different machine. Qualitative mechanisms travel; specific degree figures do not.